Compound-specific hydrogen isotope ratios of

sedimentary n-: a new palaeoclimate proxy

Dissertation

Zur Erlangung des akademischen Grades doctor rerum naturalium

(Dr. rer. nat.)

vorgelegt dem Rat der Chemisch-Geowissenschaftlichen Fakultät der Friedrich-Schiller-

Universität Jena von Diplom-Geologe Dirk Sachse

Geboren am 24.09.1974 in Halle/Saale

Gutachter:

1. PD Dr. Gerd Gleixner (Max-Planck Institut für Biogeochemie Jena)

2. Prof. Reinhard Gaupp (Friedrich-Schiller Universität Jena)

3. Prof. John M. Hayes ( Hole Oceanographic Institution, Massachustetts, U.S.A.)

Datum der öffentlichen Verteidigung: 29. Juni 2005

Index

INDEX...... 3

1. INTRODUCTION...... 5

1.1. OBJECTIVES...... 7

1.2. THESIS ORGANISATION ...... 7

2. METHODS/STRATEGIES...... 9

2.1. SAMPLE SITES AND FIELD SAMPLING...... 9

18 2.2. ANALYSIS OF WATER SAMPLES FOR δ O AND δD ...... 12

2.3. SAMPLE PREPARATION, BIOMARKER IDENTIFICATION AND QUANTIFICATION...... 12

2.4. GAS CHROMATOGRAPHY TEMPERATURE CONVERSION ISOTOPE RATIO MONITORING MASS

SPECTROMETRY (GC-TC-IRMS) FOR ANALYSIS OF δD ON THE N-ALKANES ...... 13

2.5. CALCULATION OF THE ISOTOPIC FRACTIONATION ε...... 14

3. HYDROGEN ISOTOPE RATIOS OF LACUSTRINE SEDIMENTARY N-ALKANES RECORD MODERN CLIMATE VARIABILITY...... 15

3.1. INTRODUCTION ...... 15

3.2. RESULTS AND DISCUSSION...... 16 3.2.1. n- concentrations ...... 16

3.2.3. δD and δ18O values of water...... 17

3.2.4. δD values of the n-alkanes ...... 19

3.2.4.1. δD values of the even carbon numbered short-chain n-alkanes (n-C12, n-C14, n-C16, n-C18, n-C20) and n-C13

and n-C15...... 20

3.2.4.2. δD values of the n-C17, n-C19, n-C21, n-C23 alkanes...... 22

3.2.4.3. δD values of the even carbon numbered medium to long-chain n-alkanes (n-C22, n-C24, n-C26, n-

C28, n-C30) ...... 24

3.2.4.4. δD values of the odd carbon numbered long-chain n-alkanes (n-C25 to n-C31) ...... 25

3.3. CONCLUSIONS...... 26

4. COMPOUND-SPECIFIC δD VALUES OF N-ALKANES FROM TERRESTRIAL ALONG A CLIMATIC GRADIENT – IMPLICATIONS FOR THE SEDIMENTARY BIOMARKER RECORD ...... 30

4.1. INTRODUCTION ...... 30

4.2. RESULTS AND DISCUSSION...... 31

4.2.1. n-alkane concentrations of biomass ...... 31

4.2.2. δD values of n-alkanes ...... 34 4.2.2. Hydrogen isotope fractionation between source water and biomass n-alkanes ...... 37

4.2.3. Comparison with sedimentary n-alkane distributions and δD values...... 39

4.3. CONCLUSIONS...... 42

5. SEASONAL VARIATIONS IN DECIDUOUS N-ALKANE δD AND δ 13C VALUES: IMPLICATIONS FOR THEIR USE AS A PALAEOCLIMATE PROXY ...... 44

5.1. INTRODUCTION ...... 44

5.2. MATERIALS AND METHODS ...... 45 5.2.1. Study site and sample collection ...... 45 5.2.2. Meteorological measurements at the Hainich site ...... 46 5.2.3. Modelling leaf water isotopic enrichment...... 46

5.3. RESULTS...... 48 5.3.1. n-alkane concentrations ...... 48

5.3.2. n-alkane δD and δ13C values ...... 50

5.4. DISCUSSION...... 52 5.4.1. Seasonal variations in n-alkane concentration...... 52

5.4.2. Seasonal variations in n-alkane δD and δ13C values for Maple ...... 52

5.4.3. Seasonal variations in n-alkane δD and δ13C values for leaves - Comparison of modelled leaf

water enrichment with n-alkane δD values ...... 54

5.4.4. Implications for the use of n-alkane δD values as a palaeoclimate proxy ...... 60

6. CONCLUDING REMARKS...... 61

7. SUMMARY...... 66

8. ZUSAMMENFASSUNG...... 68

ACKNOWLEDGEMENTS...... 71

9. REFERENCES...... 73

Introduction

1. Introduction

The reconstruction of past climate variability can provide valuable information on the causes, timing and magnitude of climate change. Since direct observational evidence for climatic changes is available only for the past 100 years, so-called palaeoclimate proxies (e.g. isotope ratios, data, palaeontology, sedimentological methods etc.) can provide us with a better understanding of natural climate variability and it’s forcing mechanisms. Analysis of ice-cores from the polar regions and sediments have provided valuable information on the natural climate variability throughout the geological time. The obtained data are used to evaluate climate models, which are needed to predict possible scenarios of anthropogenic climate change. However, even with the improved knowledge of the climate system components obtained over the last years, it is evident that the interactions between the biosphere, atmosphere, lithosphere and especially the hydrosphere are, despite their generally accepted importance, still poorly understood. Research on dynamics of the water cycle in the recent geological past usually focused on the oceans, because of the availability of complete sedimentary archives. Important factors influencing terrestrial ecosystems, such as atmospheric circulation, precipitation and evapotranspiration, could not be reconstructed directly because of a lack of direct proxies. Through the use of indirect proxies, such as sedimentological evidence, fossil plants, etc., some information became available, but remains scarce. It was recognised long ago that the stable isotopes of water (protium/deuterium, oxygen16/oxygen18) can be used to follow fluxes of water in the water cycle. The ratio of these isotopes in water (expressed as relative deviations from “standard mean ocean water”, SMOW, in δD and δ18O values) holds information on the water source region, temperature and evaporation (CRAIG, 1961; GAT, 1996). Tropical ice-cores are doubtlessly the best terrestrial

18 archives for water cycle reconstruction yet, because δD and δ O values can be determined directly on the ice. Data from ice-cores extended the knowledge about atmospheric dynamics in the recent geological past (THOMPSON et al., 2003). However, ice-cores are limited to the high- altitude regions, such as the Andes, the Himalayas and partly Africa (Kilimanjaro). With the global retreat of glaciers as a first effect of Global Warming (OERLEMANS, 2005), these valuable climate archives will be severely damaged and eventually be lost. δ18O values determined on carbonate sediments and/or carbonate shells from lake sediments have been shown to record the source water δ18O value and therefore variations in climate (VON GRAFENSTEIN et al., 1999). However, interpretation of these records is not straightforward, since the fractionation of oxygen isotopes during incorporation into the shells is dependent on

5 Introduction species, water temperature and the salt content of the water, which can only be estimated from other proxies for the geological past. The development of compound-specific isotope ratio measurement on individual organic compounds, first for carbon (HAYES et al., 1990) and in the late 1990ies for hydrogen

(BURGOYNE and HAYES, 1998; HILKERT et al., 1999) opened new possibilities in tracing the carbon and hydrogen sources of individual biomarker compounds. The δD values of cellulose and also bulk lipids from biomass were already proven to record the source water isotopic composition (ESTEP and HOERING, 1980; NORTHFELT et al., 1981; STERNBERG, 1988). However, cellulose is not stable over geological timescales and δD values of bulk lipids vary due to different biochemical pathways of lipid biosynthesis and associated distinct isotopic fractionations. n-Alkanes belong to one biochemical group of compounds, contain only carbon- bound hydrogen, which is non-exchangeable for temperatures up to 150°C (SCHIMMELMANN et al., 1999) and hold information on their biological source. n-Alkanes are abundant in sediments from the geological past and routinely used for palaeo-ecosystem reconstructions. These compounds are of particular interest since different classes of these can be used to distinguish terrestrial (i.e. higher plant) from aquatic (i.e. phytoplankton) production. Sessions et al. (1999) were the first to suggest, that different lipid classes have different hydrogen isotopic fractionations between lipid and source water (ε), explaining the observed variability in the bulk lipid fraction. Although only 3 different organisms were investigated an ε value of - 160‰ was observed for n-alkanes in laboratory studies. It was hypothesised that ε is independent of environmental parameters and only dependant on the biochemical pathway used for lipid synthesis. If this were the case, sedimentary n-alkane δD values could be used to reconstruct the isotopic composition of the organisms’ source water, and therefore variations in climate. In the following years applications of this new proxy for palaeoclimate reconstruction became already available with promising results over various geological timescales (ANDERSEN et al., 2001; HUANG et al., 2002; SACHSE et al., 2004a; XIE et al., 2000). However, a systematic study investigating the relationships between source water δD values and n-alkane δD values in natural ecosystems is still missing.

6 Introduction

1.1. Objectives The aim of this thesis was to present a systematic study, comparing modern sedimentary and biomass n-alkane δD values with climate, to offer a solid base for the application of this new and promising palaeoclimate proxy. Therefore sediments from 12 lakes throughout Europe following a gradient in meteoric water δD values from -120‰ in Northern Finland to -30‰ in Southern Italy, were recovered and analysed for n-alkane δD values. To understand the origin of the terrestrial biomarker record, the dominating biomass surrounding these lakes was also sampled. It is well known, that the leaf water in higher terrestrial plants, which is the source water plants use for biosynthesis, is enriched by evaporation processes in the plants leaves by 20-80‰ relative to the meteoric water (DAWSON and EHLERINGER, 1993; LEANEY et al., 1985; ZIEGLER, 1989). In temperate climates, as in Europe, the plant’s leaves will be deposited within lakes in autumn. So, if n-alkanes are produced de novo year-round in the leaf , the autumn signal would be preserved. It is known, that the chemical composition, e.g. the relative abundance of different compounds, of leaf waxes is changing over the year (GULZ, 1994), however, no such data exist for compound- specific isotope values. Consequently, in this thesis the following questions, will be addressed:

- Do n-alkane δD values from lake sediments record the modern climate variability? - Is the isotopic fractionation of hydrogen during lipid biosynthesis (ε) in natural systems independent of environmental parameters? - Is the isotopic enrichment observed in the leaf water of plants also represented in the leaf wax n-alkanes and in sedimentary n-alkanes of terrestrial origin? - Do δD values of leaf wax n-alkanes show seasonal variations?

These issues culminate in the question: How robust are sedimentary n-alkane δD values as a proxy for changes in the water cycle over geological timescales?

1.2. Thesis Organisation The thesis is organized into 9 chapters. The following chapter 2 explains the common methodologies used in all studies (measurement of δD values on individual organic compounds) as well as the sampling sites. Chapters 3 to 5 are peer-reviewed published (Chapter

3 was published in Geochimica et Cosmochimica Acta (SACHSE et al., 2004b)) or submitted (Chapter 4 and 5) articles.

7 Introduction

Chapter 3 describes the sedimentary n-alkane δD values of 12 European lakes, chapter 4 deals with the associated plant biomass n-alkanes from the lake catchments, whereas seasonal variation in leaf wax n-alkane δD values over the 2004 growing season are discussed in Chapter 5. The thesis ends with a concluding discussion (Chapter 6), where results are synthesised, a summary (Chapter 7 and 8, in German) and the references (Chapter 9).

8 Methods/Strategies

2. Methods/Strategies

2.1. Sample Sites and Field Sampling All sampled lakes (Figure 1) are small, groundwater fed lakes with a relatively small catchment area (Table 1) and low human impact except Pääjärvi Lake in Finland, which is significantly bigger and Lago di Massaciucoli, which is subject to temporal salt-water influx from the Mediterranean Sea. Six lakes are oligotrophic, 4 lakes eutrophic, one lake is mesotrophic and one is oligo/mesotrophic. For an overview of basic limnological parameters see Table 2. The sites cover a mean annual temperature range from –2.0 °C (Naimakka, northern Sweden) to 13.7 °C (Monticchio, southern Italy). Temperature and evaporation are the main factors controlling the δD value of the meteoric water, source water for the lakes (GONFIANTINI, 1986). The resulting mean annual δD value of meteoric water on the sites cover a range from –119.0 ‰ vs. VSMOW in northern Sweden/Finland (mean annual δD from 1992-1995 in Naimakka,

Figure 1: Location of the sampling sites and mean annual δD values of meteoric water over Western Europe (IAEA, 2001)

9 Methods/Strategies

Sweden) to –36.6 ‰ vs. VSMOW in southern Italy (mean annual δD from 2000 in Bari) (IAEA, 2001) (Figure 1). Sediments were sampled in August and September 2002 using a gravity corer (HTH-Teknik, Luleå, Sweden) operated from a dismountable raft. The new sediment covering roughly the last year (usually the upper 1 to 2 cm), was collected up to 4 times in the deepest part of the lake, determined from depth maps and echo-sounding. In Lake Pääjärvi, the eastern basin, whose maximum depth is 44 m, was sampled. Water temperature-, pH-, oxygen saturation and redox profiles were determined on-site using a Multiprobe (YSI Inc., Yellow Springs, Ohio, U.S.A.), ensuring comparability of the lakes. Water samples were collected using a water sampler (Hydro-Bios Apparatebau GmbH, Kiel, Germany) and a pump from the high productivity zone of the lake (usually 1-2 m). Additionally the in- and outflow of the lake, if existent, was sampled. All water samples (1l) were filtered trough a 0.45µm GF/F filter. The GF/F filters were frozen and later used for a concentration measurement (Table 2). For this purpose filters were ground and dissolved in acetone, the extinction of the extract was measured using a UV/VIS spectrometer (Helios gamma, ThermoSpectronic, Madison, WI, U.S.A.) following the procedure from Jeffrey and Humprey (1975). Phosphorus concentration was determined on-site using a field photometer (PhotoLab S12, WTW GmbH & Co. KG, Weilheim, Germany). Pooled leaf samples (about 20 shaded leaves from 2 m height) of the 2-3 dominating deciduous tree species were collected from the shores of the lakes, expect for lake Grystinge (site SOR). Additionally, two forest sites in northern Finland (SOD004 and FIN002) and one in Italy (ITA001) were sampled for leaves and stream water (representing meteoric water) if available. At the Scandinavian lakes mosses, Sphagnum species and Cladonia species make up significant portions of the vegetation, and were therefore also sampled. Since most of the lakes are situated close to a permanent research station, basic meteorological data such as temperature and precipitation data exist for the last 10 years. Long-term precipitation isotope data are taken from the IAEA GNIP database (IAEA, 2001) accessible on the internet. Additionally the Online Isotope Precipitation Calculator (OIPC, accessible at http://www.waterisotopes.org/) using the IAEA database and interpolation algorithms described in Bowen and Revenaugh (2003) was used for sites where no δD values exist. Calculated values were compared with available data and showed virtually no differences (Figure 3).

10 Methods/Strategies

Table 1: Location and basic geographic and meteorological data of the sample sites. Mean annual temperature data and mean annual precipitation data from: NAI, KEI: personal communication D. Hammarlund; SOD: Finnish Meteorological Institute (personal communication T. Laurilla); HYY: Helsinki University (personal communication T. Vesala); LAM: Lammi Biological Station (personal communication L. Arvola); SOR: wetter.com for Kopenhagen; HZM: Manderscheid station, Deutscher Wetterdienst; MAS: worldclimate.com for Pisa; MEZ: data from Valentano (RAMRATH, 1997).; LGM, LPM: Watts et al. (1996). Longitude, latitude and altitude (map datum WGS 84) were determined on-site using a handheld GPS. Water depth (e.g. of the sampling site) was determined using an echosounder; mean annual precipitation δD values were taken from the OIPC and the IAEA database (see text).

sample Lake Geographic location Altitude Mean annual Mean annual mean annual sampled code or location [m asl] temperature [°C] precipitation [mm] δD precipitation biomass

FIN002 near village of Kilpisjärvi (FIN) 69°00'16"N 20°53'52"E 496 - - -115.7 Betula pubescens NAI Oikojärvi (FIN) 68°50'55"N 21°10'50"E 463 -2.0 450 -114.9 Betula pendula -114.9 Moss -114.9 Sphagnum KEI Keitjoru (SWE) 68°40'7"N 21°30'58"E 428 -2.0 450 -114.1 Betula pubescens -114.1 Cladonia SOD003 Hirviiänkurunlampi (FIN) 67°22'44"N 26°51'3"E 178 -0.3 529 -103.9 Betula pubescens -103.9 Cladonia -103.9 Fagus sylvatica SOD007 Tunturilampi (FIN) 67°21'23"N 27°10'6"E 293 -0.3 529 -103.9 Betula pendula -103.9 Cladonia -103.9 Sphagnum SOD004 near village of Luosto (FIN) 67°10'20"N 26°51'35"E 300 - - -103.9 Betula pendula -103.9 Moss HYY Kiuvajärvi (FIN) 61°50'55"N 24°16'46"E 149 3.5 640 -92.9 Betula pendula -92.9 Moss -92.9 Sphagnum SYR Syrjänalunen (FIN) 61°11'37"N 25°8'29"E 156 - - -91.4 Myrte -91.4 Betula pendula -91.4 Moss LAM Pääjarvi (FIN) 61°3'36"N 25°5'55"E 151 3.6 619 -91.3 Betula pendula -91.3 Cladonia SOR Grystinge (DK) 55°33'11"N 11°41'42"E 30 8.6 636 -67.1 - HZM Holzmaar (GER) 50°7'10"N 6°52'44"E 436 10.5 1042 -60.1 Betula pendula -60.1 Fagus sylvatica MAS Lago di Massachiucoli (ITA) 43°50'N 10°18'49"E 18 14.1 906 -41.6 Quercus variabilis -41.6 Alnus incana -41.6 Quercus cerris ITA001 near village of Castigliano (ITA) 42°45'14"N 10°55'22"E 11 - - -38.9 Quercus robur MEZ Lago di Mezzano (ITA) 42°36'46"N 11°46'9"E 466 13.1 1030 -45.7 Quercus petraea -45.7 Carpinus betulus LGM Lago Grande di Monticchio (ITA) 40°55'58"N 15°36'10"E 674 13.7 815 -47.3 Fagus sylvatica LPM Lago Piccolo di Monticchio (ITA) 40°55'55"N 15°36'48"E 685 13.7 815 -47.3

11 Methods/Strategies

Table 2: Limnological data of the sampled lakes as determined on-site. T – temperature; DO – dissolved oxygen, TDS – total dissolved solids; chl a – chlorophyll a; PO4-P: phosphorous (as PO4 species)

sample Water Lake Catchement Secci depth water T DO saturation TDS pH chl a conc. PO4-P trophic state code depth area area surface below thermocline surface surface above thermocline [m] [km2] [km2] [m] [°C] [%] [!S] [!g/l] [!g/l]

NAI 8.4 1.5 4.5 3.65 16.5 73 27 6.82 3.6 <50 oligotrophic KEI 2.0 0.01 1.5 2.00 15.0 93 14 6.25 0.3 <50 oligotrophic SOD003 4.7 0.011 > 50 3.87 15.7 92 31 6.86 2.9 <50 oligotrophic SOD007 10.7 0.36 8.4 2.95 16.3 50 15 7.28 0.7 <50 oligotrophic HYY 13.3 0.84 30 1.56 21.1 44 28.7 6.57 5 <50 mesotrophic SYR 8.5 0.3 n.d. 4.35 20.6 92 51 7.67 0.7 <50 oligotrophic LAM 44.9 13.42 244 1.77 23.3 50 93 7.18 3.4 n.d. mesotrophic SOR 8.2 2.64 n.d 0.70 22.1 92 420 8.64 63.4 60 eutrophic HZM 24.0 0.058 2 1.40 20.3 10 246 8.5 4.9 50 eutrophic MAS 2.0 7 n.d. 0.30 25.5 n.d. 4250 8.3 7.3 <50 eutrophic MEZ 30.4 0.445 0.907 9.96 22.7 88 197 7.77 0.1 <50 oligotrophic LGM 36.7 0.405 2.37 1.40 22.5 20 440 7.25 6.8 <50 eutrophic LPM 36.0 0.08 2.37 3.50 20.6 n.d. 350 7.72 2.5 <50 oligo/meso

2.2. Analysis of water samples for δ18O and δD The isotope ratios on water were measured by on-line high-temperature (1300°C) reduction in a modified glassy carbon reactor of a high-temperature elemental analyser (TC/EA) coupled to an IRMS (DeltaplusXL, Finnigan MAT, Bremen, Germany). The average standard deviation (2σ) was 0.5 ‰ for δD values and below 0.1 ‰ for δ18O values. A detailed description of the methodology is given in Gehre et al. (2004).

2.3. Sample preparation, biomarker identification and quantification Sediment samples were ground and freeze dried. Soluble organic matter was extracted using an accelerated solvent extractor (ASE200, Dionex Corp., Sunnyvale, U.S.A.) with dichloromethane/methanol mixture (10:1) at 100°C and 2000 psi (138 bar) for 15 min in 2 cycles. Depending on the amount of organic C in the sediment samples 2-7 g of sample were used for extraction. Since biomass samples contain much higher amounts of extractable hydrocarbons than sediment samples, 1g of leaf material was sufficient for extraction. Where sufficient amount of sediment was available, the extraction procedure was performed on 2-3 independent samples. The total extract was separated on a chromatographic column made of glass (ca. 20 cm height, 2.5 cm diameter) (QVF Labortechnik GmbH, Ilmenau, Germany) filled with ca. 25 cm3

12 Methods/Strategies activated silica gel (0.040 – 0.063 mm mesh size; Merck KGaA, Darmstadt, Germany) into 3 fractions: aliphatic compounds were eluted from the column with 60 ml hexane, aromatics with 40 ml and other compounds with 40 ml methanol. The chloroform and methanol fractions were archived. Compounds of the aliphatic fraction were identified and quantified using a GC-FID (TraceGC, ThermoElectron, Rodano, Italy) by comparison to an external n- alkane standard mixture (n-C10 to n-C32). The GC was equipped with a DB5ms column (30 m, ID:0.32 mm, film thickness: 0.5 µm, Agilent, Palo Alto, U.S.A.). The PTV injector was operated in constant temperature mode at 280°C with a split ratio of 1:10. The oven was held for 2 min at 80°C then heated at 8°C/min to 320°, where the final temperature was held for 5 min. The column flow was constant at 1.8 ml/min throughout the run.

2.4. Gas Chromatography Temperature Conversion Isotope Ratio Monitoring Mass Spectrometry (GC-TC-IRMS) for analysis of δD on the n-alkanes 1 µl of the hexane dissolved aliphatic fraction was injected into a HP5890 GC (Agilent Technologies, Palo Alto, U.S.A.), equipped with a DB5ms column (30 m, ID:0.32 mm, film thickness: 0.5µm, Agilent). The injector was operated at 280°C in splitless mode. The oven was maintained for 2 min at 60°C then heated at 6°C/min to 320°C and held for 10 min at the final temperature. The column flow was held constant at 1.7 ml/min throughout the run. One fraction of the separated compounds was transferred to an ion-trap mass spectrometer (GCQ, ThermoElectron, San Jose, U.S.A.) to monitor possible co-elution of the n-alkanes with other substances. If co-elution was evident in the MS spectrum, the δD value was not used for interpretation. The other fraction was transferred to a high-temperature conversion oven operated at 1425°C (BURGOYNE and HAYES, 1998; HILKERT et al., 1999) and quantitatively converted to H2, which was introduced into an isotope ratio mass spectrometer (IRMS) (DeltaplusXL, Finnigan MAT, Bremen, Germany) for compound-specific analysis of δD values. Each sample was independently measured 3 times. All δD values were normalized to the

VSMOW scale using a mixture of n-alkanes (n-C10 to n-C32). The δD values of the n-alkanes in the standard mixture were calibrated against international reference substances (NBS-22, IAEA- OH22). Individual n-alkanes were analysed for their δD values using the offline TC/EA (Thermal Conversion/Elemental Analyser) technique, a detailed description of the calibration and standardization procedure can be found in Sachse (2002). After the measurement of no more than 2 samples (6 GC runs), the standard mixture was

+ measured independently 3 times. If necessary a drift correction was applied. The reaction of H2

13 Methods/Strategies

+ ions with H2 molecules in the ion source of the IRMS results in the formation of H3 ions, + which also contribute to the measured mass 3. Since the amount of H3 ions formed in the source is a function of the gas pressure, a correction can be applied by determining the so-called + + H3 factor (WERNER and BRAND, 2001). A constant H3 factor therefore indicates stable ion + source conditions. The H3 factor is usually determined once a day. Achieved precision can be expressed as the average standard deviation for all measured samples and standards (3 individual measurements per sample or standard) and is given in Table 3 for the 3 measurement

+ campaigns. The H3 factor was constant during the measurement periods (Table 3).

+ Table 3: Standard deviation (2σ stdv) and H3 factor of all measured samples and standard mixtures.

+ Standard Total No. No. of Samples Total no. No. No. of H3 No. stdv mixture no. of of peaks stdv of peaks of peaks per factor stdv peaks runs per runs run run Campaign 1 (Chapter 3) 4‰ 3520 160 22 7‰ 2700 180 15 5.9 23 0.3 Campaign 2 (Chapter 4) 3‰ 437 23 19 4‰ 114 60 1-3 3.9 11 0.1 Campaign 3 (Chapter 5) 4‰ 1620 90 18 4‰ 303 93 1-8 4.0 14 0.1

The same instrumental setup as for δD measurement was used for carbon isotope analysis on the leaf lipids (Chapter 5). Here the GC was coupled to the IRMS via a GC Combustion III interface, with the oxidation oven operated at 960°C. The δ13C values were normalized to the VPDB scale using the same n-alkane mixture as for hydrogen, with predetermined TC/EA δ13C values. Average standard deviation for all peaks was 0.3‰ for the samples (n=114, 60 runs with 1-3 peaks each, depending on sample), 0.4‰ for the standard mixtures (n=323, 17 runs

with 9 peaks each, n-C14 to n-C22).

2.5. Calculation of the isotopic fractionation ε The isotopic difference between the δD value of the lake water and the δD value of the n- alkanes was calculated using equation 1.

 +1000  =1000⋅ δDalkane −1 Equation 1 ε alkane / water +1000  δDwater 

14 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability

3. Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability

3.1. Introduction Compound specific hydrogen isotope ratios of organic compounds are emerging as a new palaeoclimatic and palaeohydrological proxy (ANDERSEN et al., 2001; HUANG et al., 2002;

SACHSE et al., 2004a; SAUER et al., 2001; YANG and HUANG, 2003). The use of this technique on compounds originating from specific groups of organisms, so called biomarkers, opens new perspectives to trace changes in various compartments of the water cycle and hence climate variability in the geological past. The hydrogen isotope composition of environmental water is dependent on climatic conditions such as temperature, evaporation, precipitation and others

(CRAIG, 1961; CRAIG and GORDON, 1965; GAT, 1996; GONFIANTINI, 1986). Current knowledge suggests that the fractionation of H isotopes in biosynthesis is constant and mostly controlled by the biochemical pathway used (SESSIONS et al., 1999). Therefore δD ratios of biomarkers have the potential to record changes in the isotopic composition of the H source. This is well known for δD values from cellulose of tree rings (EPSTEIN et al., 1976; FENG and EPSTEIN, 1995; STERNBERG, 1988; YAPP and EPSTEIN, 1982) and also for bulk plant lipids (STERNBERG,

1988). For n-alkanes from a fractionation (εn-alkane/water) of –160 ‰ and for sterols –201‰ has been reported (SAUER et al., 2001; SESSIONS et al., 1999). Moreover, n-alkanes from terrestrial higher plants are more enriched in deuterium, resulting in an εn-alkane/water value of about –117‰ for C3 plants (CHIKARAISHI and NARAOKA, 2003; ESTEP and HOERING, 1980;

STERNBERG et al., 1984), since leaf water in plants, the H source for biosynthesis, is usually enriched by 20 to 80 ‰ relative to soil water (ZIEGLER, 1989).

Lake sediments are excellent climate archives (e.g. ALLEN et al., 1999), recording regional climate changes on the continents and are therefore the prime target for biomarker δD analysis. Evidentially, palmitic acid extracted from recent lake sediments throughout North America records the δD of the source water (HUANG et al., 2002). Other more specific biomarkers such as several sterols also record the δD value of the source water (SAUER et al., 2001). However, as carboxylic acids are only minor compounds in older sediments, those biomarkers are not always suitable for palaeoclimatic reconstruction. Therefore, n-alkanes, which are present in considerable amounts even in palaeozoic sediments, do not contain exchangeable H

(SCHIMMELMANN et al., 1999) and are relatively easy to extract and purify, should be reliable compounds in terms of δD analysis. The n-alkanes from recent sediments cover a wider range of δD values (SAUER et al., 2001), due to different biological sources. The occurrence of the short chain even-carbon numbered n-

15 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability

C12 to n-C22 alkanes in sediments from different environments is generally attributed to bacteria (GRIMALT and ALBAIGES, 1987; HAN and CALVIN, 1969). Short-chained n-alkanes, especially n-C17, are generally considered as indicators for input by algae and photosynthetic bacteria (CRANWELL et al., 1987; HAN and CALVIN, 1969; HAN et al., 1968; MEYERS, 2003), n-alkanes with 21, 23 and/or 25 carbon atoms are derived from submerged aquatic plants (FICKEN et al.,

2000), whereas the n-C25 to n-C31 alkanes originate from the leaf waxes of land plants

(CRANWELL et al., 1987; EGLINTON and HAMILTON, 1967). The n-C23, n-C25 and n-C31 alkanes can also be derived from sphagnum species (BAAS et al., 2000). Consequently, the δD values of different n-alkanes should yield information on the origin of hydrogen in the bacteria, algae and plants. First reconstructions of palaeohydrological conditions using δD values from sedimentary n- alkanes of various ages up to the Permian show promising results (ANDERSEN et al., 2001;

SACHSE et al., 2004a; SAUER et al., 2001; XIE et al., 2000). So far a systematic approach comparing recent sedimentary n-alkane δD values with climate data is still lacking. Therefore, we sampled surface sediments from 12 lakes along a N-S climatic gradient from Northern Finland to Southern Italy to test if the climatic gradient is recorded in n-alkane δD values and to present a solid base for the application of n-alkane δD values as a new palaeohydrological proxy.

3.2. Results and Discussion

3.2.1. n-alkane concentrations In general, the total concentration of extractable n-alkanes is slightly higher in the lakes from southern Europe (up to 0.5 mg/g TOC) compared to the northern lakes (0.2 mg/g TOC as a minimum), indicating a higher organic input in the south (Figure 2). The concentration of the short-chain n-alkanes n-C17 and n-C19 of aquatic origin is highest in the eutrophic lakes (up to

0.4 mg/g TOC for n-C17 in HZM). Most prominent n-alkanes in nearly all lake sediments are the long chain n-alkanes n-C25 to n-C31 of terrestrial origin, expressed through the high average chain length (ACL) values varying between 21.2 and 27.6 (Figure 2). A clear odd over even carbon number predominance was found in all sediments with carbon predominance indices (CPI) between 1.4 and 6.6 (Figure 2), typical for a biological origin. The CPI increases slightly from north to south, suggesting together with the longer vegetation period, higher input of organic material and higher productivity in the southern lakes, although dilution with other sedimentary components and different preservation regimes may also influence the TOC contents.

16 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability

Figure 2: Distribution, CPI (Carbon Preference Index = Σodd Cn / Σeven Cn) and ACL (Average Chain Length = (ΣCn*n)/ ΣCn) indices and total concentration (HCtot) of the n-alkanes from 9 of the sampled lakes. Note that the scale of the y-axis (concentration) is doubled for the southern lakes (HZM, MAS, LGM, LPM) because of higher concentration of n-alkanes.

3.2.3. δD and δ 18O values of water

18 The δD and δ O values of the lake inflow water, if an inflow is present, are up to 14 ‰ heavier for δD and up to 1.5 ‰ heavier for δ18O relative to the calculated OIPC values (Fig. 3).

17 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability

Considering the confidence interval of the calculated OIPC data, the inflow water δ values are within the 95% limit or slightly heavier. Due to evaporation and temperature effects a heavier summer δ value would be expected. The northernmost lake NAI represents a notable exception, 18 since measured δD values are 21 ‰ and δ O values 3.4‰ heavier than the calculated meteoric water. Since there is no dense vegetation cover around the small stream and summer 2002 was exceptionally warm and dry in Scandinavia, evaporation already affects the contributing stream. However, the isotopic data for the inflowing waters from the 7 lakes where those data exist, suggest, that the main water source is the meteoric water. Evaporation only seems to be of minor influence.

Table 4: Isotopic composition of meteoric water at the lake sites, taken from the IAEA GNIP database (IAEA, 2001) for the following sites: 1)Naimakka (1992-1995) about 1km from KEI site, 2)Taastrup (1965-1971), 3)Koblenz (1974-1975) 4)Pisa (1992-1995), 5)Ancona (2000), 6)Bari (9-12/2000). Note: except for Naimakka (KEI, NAI site), the IAEA sites are not directly at the sampling site, the differences in IAEA and OIPC data are due to the different geography (altitude, distance to the sea etc.). OIPC δD and δ18O meteoric water data and 95% confidence interval (CI) were calculated using the Online Isotopes in Precipitation Calculator (OIPC) using the interpolation algorithms described in Bowen and Revenaugh (2003). Lake and inflow water δD and δ18O values were measured. All δ values given in ‰ vs. VSMOW.

sample δD meteoric δD meteoric 95% δ18Ο meteoric 95% δD inflow δ18Ο inflow δD lake water δ18Ο lake water code water (IAEA) water (OIPC) CI water (OIPC) CI stdv (2σ) stdv (2σ) stdv (2σ) stdv (2σ)

NAI -119.0 1) -114.9 9.2 -15.43 1.12 -93.7 0.4 -12.04 0.16 -94.9 0.6 -12.5 0.0 KEI -119.0 1) -114.1 8.6 -15.32 1.06 -100.1 0.1 -13.75 0.07 -96.6 0.6 -13.62 0.04 SOD003 n.a. -103.9 8.0 -14.04 1.17 -109.6 1.0 -14.55 0.01 -103.7 0.2 -13.92 0.02 SOD007 n.a. -105.2 8.1 -14.23 1.19 -108.8 0.4 -14.98 0.05 n.d. n.d. -12.29 0.05 HYY n.a. -92.9 7.9 -12.85 1.12 - - - - -81.1 0.6 -10.59 0.00 SYR n.a. -91.4 7.0 -12.38 1.02 - - - - -86.5 0.3 -11.95 0.06 LAM n.a. -91.3 7.0 -12.36 1.02 - - - - -73.7 0.0 -9.55 0.14 SOR -70.7 2) -67.1 4.5 -9.45 0.61 -55.2 0.5 -7.79 0.05 -46.4 0.7 -6.40 0.08 HZM -52.1 3) -60.1 3.1 -8.67 0.51 -54.5 0.7 -8.48 0.09 -47.1 0.6 -6.44 0.07 MAS -47.0 4) -41.6 6.4 -6.47 0.71 - - - - -6.3 0.6 -0.24 0.05 MEZ -40.1 5) -45.7 6.3 -7.04 0.69 -36.8 0.3 -6.29 0.17 -3.5 0.6 0.65 0.11 LGM -36.6 6) -47.3 5.2 -7.23 0.56 - - - - -13.5 0.2 -0.92 0.10 LPM -36.6 6) -47.3 5.2 -7.23 0.56 - - - - -32.6 0.6 -3.99 0.09

δD and δ18O values of the lake water from the high productivity zone (in general at 1-2m depth) are up to 20 ‰ heavier for hydrogen and up to 3 ‰ heavier for oxygen than the calculated OIPC values, except the three Italian lakes MAS, MEZ and LGM (Fig. 3). Nearly half of the variation still lies within the 95% confidence interval of the calculated OIPC data (Table 4). Furthermore, it must be acknowledged that the isotope ratios of the lake water are only a one- point measurement in summer, when evaporation very likely will enrich deuterium in the water.

18 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability

However, since the organisms in the water column will incorporate hydrogen throughout the growing season and probably show most productivity in spring and late summer, the summer δD water value will be heavier than the average δD water value for the growing season. We therefore assume the OIPC data to be a mean annual value also valid for the lake water. For the Italian lakes MEZ and LGM a stronger evaporation effect is visible. The lake water is 33 to 35 ‰ heavier in δD than the calculated precipitation value (Fig. 3). Lake MAS, also in Italy, is a special case, because it lies within several kilometres of the Mediterranean coast and temporal influx from isotopically heavier seawater occurs. The isotopic ratios of these three Italian lakes do not plot on the Global Meteoric Water Line, but on an evaporation line, whereas all other lake waters do. In the case of the three Italian lakes an underestimation of the isotopic fractionation ε would be the case, if we rely on the OIPC data. Despite this deviation in 3 southern lakes, the OIPC data have been used as the hypothetical mean annual δD composition of the lake water.

Figure 3: Comparison between lake water δD values (open circles: high-productivity zone, usually in 1-2 m depth; filled circles: inflows) and meteoric water δD values, calculated using OIPC (see text). Dotted lines show the enrichment/depletion of inflow water vs. lake water. Error bars for the δD of OIPC data constitute the calculated 95% confidence interval and are also applicable for the inflow data. Error bars (standard deviation) for δD lake water lie within the points.

3.2.4. δD values of the n-alkanes In general the δD values of n-alkanes from one sample can differ by more than 100‰. The main reasons for this variation are different biological sources of these n-alkanes, using different water sources for biosynthesis. Using δD values, four main n-alkane groups can be distinguished:

19 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability

1) Even carbon numbered short-chain n-alkanes (n-C12, n-C14, n-C16, n-C18, n-C20) and

n-C13 and n-C15

2) Odd carbon numbered n-alkanes of medium chain length (n-C17, n-C19, n-C21, n-

C23)

3) Even carbon numbered medium to long-chain n-alkanes (n-C22, n-C24, n-C26, n-C28,

n-C30)

4) Odd carbon numbered long-chain n-alkanes (n-C25 to n-C31)

In the following we will discuss the variations of the δD values of those n-alkane groups along the transect in detail.

3.2.4.1. δD values of the even carbon numbered short-chain n-alkanes (n-

C12, n-C14, n-C16, n-C18, n-C20) and n-C13 and n-C15

δD values for n-C16 and n-C18 could be determined on only 5 lake samples along the transect due to overall low concentration of these substances (Table 5). They usually are the heaviest measured δD values of the sample. Interestingly, δD values of the compounds are lighter in the south than in the north. The δD value for n-C18 is significantly lighter than the other alkanes in the sediments from SOR and LGM sites, probably indicating a different H source for n-C18 in these two eutrophic lakes. δD values from n-C16 and n-C18 correlate positively with latitude and negatively with mean annual temperature and meteoric water δD values (Table 5, Figure 4), although the number of samples is limited. The other short-chained substances seem to follow this trend, but statistical evaluation is not possible due to the limited amount of data. εsubstance/water is increasing from the north to the south for those short-chained n-alkanes. εC16/w and εC18/w show a statistically significant inverse relationship with meteoric water δD and δD lake water values and correlate with temperature (Table 6, Figure 5). This is surprising since n-C16 and n-

C18 are together with n-C17 widely used as indicators for algal input. However, both n-alkane

δD values significantly differ from n-C17 δD values.

Our results suggest that δD values of n-C16 and n-C18 are independent of the meteoric water δD value and show an inverse relationship with mean annual temperature. The fractionation between water and n-C16 and n-C18 appears to be larger in areas with higher temperatures. To our knowledge such a relationship has not been reported until now.

Such large differences between the δD values of n-C16, n-C18 and n-C17 have not been observed in palaeozoic samples or oils (LI et al., 2001). Several explanations might be considered:

20 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability

1) Dependence of εsubstance/water on growth rate. Since ε is smaller in the north (shorter vegetation period, less nutrient supply) than in the south for these possibly non-photosynthetic bacteria derived compounds, a fractionation

dependent on growth rate is possible. Sessions et al. (1999) observed a wider range of ε values for lipids in actively growing organisms than in plants. However, growth-rate

dependency of ε for fatty acids of a methane-oxidizing bacteria could not be confirmed (SESSIONS et al., 2002).

2) Different sources for these compounds in the north and in the south. Since εsubstance/water

values in the south are closer to ε values of n-C17, they could share the same algal source in the southern lakes. The high δD values in the north could be related to methane- oxidizing bacteria, since Sessions et al. (2002) predict δD values of fatty acids produced by methanotrophs to be between –50 to –170‰. However, it is unlikely that in the sediments of the oligotropohic Scandinavian lakes significant amounts of methane are produced. Another possible source for the isotopically light compounds in the north could be fossil hydrogen, possibly introduced by petroleum contamination. But contamination seems unlikely for the Scandinavian lakes (NAI, SOD), since population density is very low and most lakes are several kilometres away from rarely used roads. It is also unlikely for Lake Grystinge (SOR), used for drinking water storage with restricted access. Also petroleum contribution should not only affect the short-chained alkanes, but all alkanes, which is not evident. Fossil oil or shale samples lack the clear odd over even carbon number predomination expressed in the high CPI indices for the analysed samples. It is unlikely that transport by wind over long distances accumulates significant amounts of those alkanes in the most remote areas, but not in more densely populated areas of southern Italy. The only lakes where motorboat traffic was allowed (LAM and MAS) do not contain sufficient amounts of short-chain alkanes and do not show lower than average CPI indices. However, a possible mixing of petroleum derived alkanes with biologically derived alkanes cannot be ruled out, a way to test this hypothesis would be compound-specific radiocarbon analysis to determine the amount of fossil (“dead”) carbon.

Further research is necessary to clarify the origin of H in the short-chained n-alkanes with even carbon numbers and if different fractionations of the hydrogen isotopes occur in heterotrophs compared to autotrophic organisms.

21 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability

Figure 4: δD values of the n-C16, n-C17, n-C18, n-C21, n-C25, n-C27, n-C29 and n-C31 alkane vs. the meteoric water δD value (calculated using OIPC, (BOWEN and REVENAUGH, 2003)). Dotted lines show the 95% confidence interval of the regression line.

3.2.4.2. δD values of the n-C17, n-C19, n-C21, n-C23 alkanes

These substances, especially n-C17 are of particular interest since they are widely used as a palaeoclimatic proxy. δD values of the n-C17 alkane were determined in the samples from 8

22 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability

lakes. The other samples either contained no n-C17 or the substance could not properly be separated on the GC column. However, n-C17 shows a clear negative correlation with latitude

(Table 5, Figure 6) and a positive one with mean annual temperature. δD values of n-C17 strongly correlate to the lake water δD value and to the δD value of meteoric water with an intercept of –156.5 (Figure 4) suggesting that the meteoric water is the H source for n-C17.

εC17/W values from the lakes range from –134 to – 174 ‰ with the mean at –157 ‰ (Table 6), about the same value as observed in laboratory experiments (SESSIONS et al., 1999). εC17/W is constant along the N-S transect (Table 6), covering different climates and lakes of different trophic state, supporting the hypothesis that the fractionation of hydrogen isotopes during incorporation into organic matter is only dependent on the biochemical pathway used and does not depend on environmental parameters. These results demonstrate that the n-C17 alkane excellently records the lake water δD value and can be used to reconstruct the δD value of lake water (see also Figure 6). n-C21 and n-C23 are considered to be derived from aquatic plants living in the water (FICKEN et al., 2000). δD values are often, but not always similar to n-C17 with more scatter (Table 5).

They are lighter in the north than in the south, following the same trend as n-C17, but, except for n-C19 (although just 3 samples provided enough substance and no co-elution), do not correlate significantly with latitude, temperature or water δD values. This might be due to the higher standard deviations of δD values of n-C19, n-C21 and n-C23 relative to n- C17, since they often sit on a hump of unresolved compounds in the chromatogram. Different water plants use water from different depths, characterized by different δD values, or they might have different evapotranspiration rates. n-C21 and n-C23 δD values seem to record the isotopic composition of lake water, but for reconstruction of the lake water δD value n-C17 is much more recommended.

23 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability

Figure 5: εsubstance/water values calculated for the n-C16, n-C17, n-C18, n-C21, n-C25, n-C27 and the n-C29 alkanes plotted vs. mean annual temperature on the lake sites. Note: only the relationships for n-C16 and n-C18 are statistically significant (n-C16: p=0.002, r=- 0.99; n-C18: p=0.006, r= -0.90).

3.2.4.3. δD values of the even carbon numbered medium to long-chain n-alkanes

(n-C22, n-C24, n-C26, n-C28, n-C30) Since the substances are present in very low concentrations, only a few samples yielded enough substance to measure δD values. Interpretation of their origin is therefore speculative. However,

24 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability the even carbon numbered substances seem to be related in δD values to their following odd carbon numbered neighbours, being in general some per mil heavier.

δD values of n-C22 were only measured in 3 samples (KEI, SOD007 and SOR) and seem to be related to n-C21 and n-C23, usually being several per mil heavier (Table 4). n-C24 (only measured in KEI and SOR) is about 20‰ heavier than n-C25. n-C26, n-C28 and n-

C30 are also 10-20 ‰ heavier than n-C27, n-C29 and n-C31 in the same sample.

Overall it seems that the source of n-C22 is the same source as for n-C21 and n-C23, and n-C24 to n-C30 share the same source as their odd carbon numbered neighbours n-C25 to n-C31, but involving some enrichment mechanism, resulting in 10 to 20 ‰ heavier δD values. Since this enrichment is not reported in plants (CHIKARAISHI and NARAOKA, 2003), it is likely that these alkanes are produced during microbial induced degradation or recycling processes of the odd- carbon numbered alkanes taking place in the sediment. However, it is also possible, that a different biological origin, for instance grass, which would be subject to different amounts of evapotranspiration than , is the reason for this slight enrichment.

3.2.4.4. δD values of the odd carbon numbered long-chain n-alkanes (n-C25 to n-

C31) These four substances, commonly considered as indicators for higher terrestrial plant input, are the most abundant aliphatic compounds in all lake sediments sampled here, surpassed only by n-C17 in some samples. All of these biomarkers show similar δD values in a given sample. They negatively correlate with latitude and positively with mean annual temperature and δD values of meteoric water and lake water (Table 5, Figure 4). Substances are isotopically lighter in the north and about 70 ‰ heavier in the south (Figure 6).

εC29/W ranges for all 12 lakes between –120 to –147 ‰ with the mean at –130 ‰ (Table 6). This is 27 ‰ heavier than the average εC17/W. Several studies have confirmed that leaf water in plants is enriched by 20 to 80 ‰ relative to soil water due to evapotranspiration processes in the leaf

(ZIEGLER, 1989). Furthermore Chikaraishi & Naraoka (2003) report an εwater value for the long- chain n-alkanes from several C3 plants of –117 ‰, consistent with our data. The long-chained odd carbon numbered n-alkanes record the meteoric water δD value. They are in general 30 ‰ heavier than n-C17 (Figure 6), due to evaporation processes in the leaves of the plants. If we consider that the aquatic and terrestrial n-alkanes share the same H source, meteoric water, the difference between δD values of both substances could serve as a proxy for evapotranspiration of the lake ecosystem. The mean difference between the terrestrial and the aquatic δD values is highest for the three largest lakes MAS, LAM and NAI in terms of lake

25 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability area and catchment area with values between 40 and 70 ‰. In these lakes higher input of allochtonous biomass from distant areas with different isotopic composition of meteoric water can lead to higher differences of both δD values. The difference ranges around 10 to 20 ‰ for the other lakes, explainable by the enrichment of leaf water due to evaporation processes in the leaf. For the lakes LGM and LPM, which are at the same site and should experience the same amount of evapotranspiration, it is virtually identical (about 9‰). However, since we have no evapotranspiration measurements from the sites, we cannot completely explain the differences between the δD values of the terrestrial vs. aquatic substances and confirm that it is indeed a marker of the evapotranspiration of the lake ecosystem. n-alkanes originate from the epicuticular waxes of leaves, developed shortly after the unfolding of the leaf in spring (GULZ, 1994), when evapotranspiration is not yet very strong. However, the n-alkane composition of the wax is changing throughout the year, differing from tree species (PIASENTIER et al., 2000). If most of the n-alkanes are produced in spring, we would not see a stronger enrichment relative to the aquatic n-alkanes for sites with higher evapotranspiration in summer. This would explain the rather constant difference of 30 ‰ along the transect.

Figure 6: Variation of meteoric water δD values, δD values and the isotopic fractionation ε for the aquatic n-C17 and the terrestrial n-C29 alkane along the N-S European transect

3.3. Conclusions Hydrogen isotope ratios of recent lacustrine sedimentary n-alkanes clearly record the isotopic composition of the meteoric source water in lakes from different climates. In particular n-C17 and n-C25, n-C27, n-C29 and n-C31 can be used to reconstruct the source water δD value. The

26 Hydrogen isotope ratios of lacustrine sedimentary n-alkanes record modern climate variability

constant fractionation εC17/W of about –157 ‰ along the transect supports the hypothesis fractionation of H during biosynthesis is independent of environmental parameters.

The enrichment of about 30 ‰ of the terrestrial long-chain n-alkanes to the aquatic n-C17 is likely due to evaporation processes in plant leaves. The variation in the difference between terrestrial and aquatic n-alkanes might be influenced by input of allochtonous biomass of distant origin on the sites. Since it is not clear if the leaf wax n-alkanes are synthesized de novo year- round, further investigation involving evapotranspiration measurements and study of possible seasonal variations in leaf wax n-alkane δD values will clarify if this difference can serve as a valuable proxy for the evapotranspiration of the lake ecosystem.

The hydrogen source of the short-chain n-alkanes n-C12 to n-C20 (excluding n-C17 and n-C19) is probably not the meteoric water pool. Non-photosynthetic bacteria synthesizing these compounds or a contribution of fossil hydrogen (petroleum contamination) might be responsible. Further research on the differences of H isotope fractionation in autotrophic vs. heterotrophic organisms is necessary. Compound-specific radiocarbon measurements on the short-chained alkanes could help clarify their origin. Our results demonstrate that hydrogen isotope ratios of recent sedimentary n-alkanes excellently record the δD value of the lake water (Figure 6) and hence enable climatic reconstruction. Moreover, the use of terrestrial and aquatic compounds might be used as a proxy for evapotranspiration of the lake environment. n-alkanes are abundant in sediments and relatively easy to purify, so the application of compound-specific hydrogen isotope measurements to samples from the geological past opens new possibilities for the reconstruction of changes in the water cycle.

27

Table 5: δD values and statistical data (p-value and the correlation coefficient r) on the measured n-alkanes. Bold values indicate a significant correlation within the 95% confidence interval (p < 0.05). SD: standard deviation (2σ) H y

sample n-C12 SD n-C13 SD n-C14 SD n-C15 SD n-C16 SD n-C17 SD n-C18 SD n-C19 SD n-C20 SD n-C21 SD n-C22 SD n-C23 SD n-C24 SD n-C25 SD n-C26 SD n-C27 SD n-C28 SD n-C29 SD n-C30 SD n-C31 SD d r

code [‰ vs. VSMOW] o g e n

NAI -146 12 -120 1 -114 10 -259 6 -131 2 -209 6 -221 2 i s

KEI -245 4 -231 11 -234 2 -214 14 -234 4 -222 n.d. -229 5 -230 5 -230 4 o t

SOD003 -142 4 -122 3 -100 3 -142 1 -235 7 -210 11 -215 9 o SOD007 -102 16 -153 12 -125 6 -268 9 -268 5 -220 3 -227 5 p e

HYY -229 6 -135 13 -244 8 -218 2 -206 3 -202 1 -205 4 -209 3 r a

SYR -207 1 -202 1 -207 2 t i LAM -272 12 -205 4 -213 6 -205 4 o s

SOR -148 n.d. -219 1 -238 4 -241 2 -145 5 -213 2 -217 4 -177 8 -193 1 -177 2 -183 4 -201 2 -181 5 -208 0 o f

HZM -138 2 -140 10 -126 10 -148 15 -170 5 -208 0 -158 14 -225 6 -180 8 -184 10 -198 1 -198 5 l a

MAS -208 3 -201 7 -235 6 7 -129 17 -159 7 c u

MEZ -198 0 -164 11 -153 13 -163 4 -167 3 s t

LGM -130 2 -175 0 -245 10 -201 5 -194 5 -180 3 -185 16 r i LPM -180 5 -168 5 -177 2 -170 3 -169 2 n e

s e

correlation vs. p r p r p r p r p r p r p r p r p r p r p r p r p r p r p r p r p r p r p r p r d i

latitude 0.26 -0.92 - 0.2 0.95 - 0.006 0.97 0.002 -0.93 0.05 0.76 0.11 -0.97 - - 0.08 -0.76 0.54 -0.66 0.05 -0.70 - - 0.000074 -0.97 - - 0.001 -0.87 - - 0.000001 -0.95 - - 0.006 -0.97 m e n t

mean annual temperature 0.22 0.94 - 0.16 -0.97 - 0.015 -0.95 0.004 0.91 0.04 -0.78 0.04 0.99 - - 0.12 0.70 0.59 0.61 0.07 0.66 - - 0.00008 0.97 - - 0.0005 0.89 - - 0.00001 0.94 - - 0.01 0.96 a r y

δD precipitation (OIPC) 0.18 0.96 - 0.12 -0.98 - 0.023 -0.93 0.007 0.89 0.03 -0.8 0.2 0.96 - - 0.15 0.74 - - 0.50 0.71 - - 0.00006 0.97 - - 0.0003 0.9 - - 0.000001 0.95 - - 0.016 0.94 n - a l D lake water k δ 0.39 0.82 - 0.33 -0.87 - 0.048 -0.95 0.015 0.85 0.03 -0.85 0.11 0.98 - - 0.23 0.58 - - 0.11 0.66 - - 0.0003 0.95 - - 0.0005 0.91 - - 0.000010 0.93 - - 0.02 0.92 a n e

s

r e c o r

d

m o

d e r n

c l

i m

a t e

v a

r i a b i l i

t y

28

28

Table 6: εsubstance/water values and 2σ standard deviation (SD) for each site and mean values along the transect. Bold values indicate constant εsubstance/water values along H

the transect y d r

o g e n

i s

o t o p e

r a t i o

sample ε substance/meteoric water s

o

code C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 C23 C24 C25 C26 C27 C28 C29 C30 C31 f

l a

[‰ vs. VSMOW] c u s t r i NAI -35 -6 1 -162 -18 -107 -120 n e

s

KEI -148 -132 -136 -113 -135 -122 -130 -130 -131 e d i

SOD003 -43 -21 4 -42 -147 -118 -124 m e

SOD007 3 -54 -22 -182 -182 -128 -136 n t a

HYY -150 -47 -166 -138 -125 -120 -124 -128 r y

SYR -128 -122 -127 n - a l

LAM -199 -125 -134 -125 k a n

SOR -86 -163 -183 -187 -84 -156 -161 -118 -135 -118 -124 -143 -122 -151 e s

HZM -83 -85 -70 -94 -117 -157 -104 -175 -128 -132 -147 -147 r e c

MAS -174 -167 -202 -91 -122 o r d

MEZ -160 -124 -113 -122 -128 m

LGM -87 -134 -207 -162 -154 -140 -145 o d e

LPM -139 -126 -137 -129 -128 r n

c l i m a t

mean -54 -85 -32 -90 -39 -157 -94 -172 -53 -161 -158 -153 -115 -126 -120 -122 -126 -130 -122 -141 e

v

SD 26 34 5 58 12 74 13 44 22 25 28 3 7 2 14 3 8 10 a r i a b i l i t y

2 9

29 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient

4. Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient – Implications for the sedimentary biomarker record

4.1. Introduction The variation of the stable isotope ratios of hydrogen in meteoric water are caused by a variety of climatic parameters, such as temperature, evaporation and water vapour source (CRAIG, 1961). Since meteoric water is the only hydrogen source for most of the organisms on earth, the potential of organic matter to record this climate signal has been a focus of research for some time (EPSTEIN et al., 1976). Especially plant cellulose was found to record the source water signal (FENG and EPSTEIN, 1995). Unfortunately, cellulose is not stable over geological time scales and therefore not abundant in the sedimentary record. Bulk lipids are geologically stable biomarkers, which are also known to record the source water δD value (STERNBERG, 1988). However, distinct isotopic fractionations during biosynthesis of various lipids lead to variations in bulk lipid δD values, which are not related to isotopic variations in source water. The development of compound-specific hydrogen isotope analysis in the last years overcomes this problem and permits the measurement of δD values on individual organic compounds. In sediments from the geological past n-alkanes are among the most abundant lipids, since they are very stable compounds. n-Alkanes are relatively easy to extract and purify and all hydrogen atoms in n-alkanes are carbon-bound and therefore non-exchangeable for temperatures up to

150°C (SCHIMMELMANN et al., 1999). Furthermore, n-alkanes originate from different biological sources, n-alkanes with 17 and 19 carbon atoms are derived from algae, n-alkanes with 23 carbon atoms can be produced by submerged aquatic plants (FICKEN et al., 2000) and sphagnum species (BAAS et al., 2000), which also produce n-C25. The n-alkanes with 25 to 31 carbon atoms, but especially n-C27 and n-C29 are produced in the leaf waxes of terrestrial higher plants (EGLINTON and HAMILTON, 1967). Very high amounts of n-alkanes are present in deciduous tree leaves, compared to significantly lower concentrations in coniferous tree needles.

The n-C29 and especially n-C31 are also major constituents of grasses (MAFFEI, 1996). Sessions et al. (1999) have demonstrated in laboratory experiments that the source water δD value is recorded in the organisms’ n-alkanes. In the previous chapter 3 (SACHSE et al., 2004b), it is shown that aquatic and terrestrial n-alkanes from recent lake sediments along a N-S European climatic gradient record the meteoric water isotope composition in natural ecosystems. The observed fractionation between source water and n-alkanes was constant at -157‰ for the

30 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient aquatic substances, which is in agreement with results from laboratory studies (SESSIONS et al., 1999). Sachse et al. (2004b) found that the terrestrial n-alkanes are enriched between 10 and 60‰ relative to the aquatic biomarkers. This is in agreement with previous work, where hydrogen isotope fractionations between source water and n-alkanes (εw/a) of about -117±27‰ for C3 plants have been reported (CHIKARAISHI and NARAOKA, 2003). However, the observed isotopic difference between higher terrestrial plants and aquatic organisms does not originate from a different biosynthetic pathway. The leaf water of plants, which is the hydrogen source for all biosynthetic products, is subject to evapotranspiration, leading to an enrichment between

20 to 80‰ (LEANEY et al., 1985). Therefore, two isotopically distinct source waters used for the biosynthesis of terrestrial and aquatic biomarkers are the reason for the observed differences. Sachse et al. (2004) have argued that the difference between aquatic and terrestrial biomarker δD values in small, closed lake systems, which are fed by the meteoric water, could serve as a proxy for the evapotranspiration of the lake ecosystem. However, the nature of the terrestrial n- alkane signal in lake sediments represents a mixture of different plant species, in temperate climates most leaves will be deposited in the lake in autumn. Therefore the n-alkane hydrogen isotope composition of the dominating mostly deciduous tree vegetation around the lakes was investigated and compared with the sedimentary biomarker record investigated in the previous study (SACHSE et al., 2004b). The n-alkane composition and δD values of 31 plant biomass samples (Betula, Fagus, Quercus, Alnus, Carpinus and Myrtus as well as non stomata containing Sphagnum, Cladonia and Moss) from 14 sites along a climatic gradient from Northern Finland to Southern Italy has been analysed (Table 1). 11 of the sites were previously investigated for the n-alkane δD values in adjacent lake sediments (see chapter 3) and are compared to the leaf wax n-alkane δD values. Variability among different species and different n-alkanes will be explored. Conclusions are drawn regarding the use of the terrestrial n-alkane δD signal as a palaeoclimate parameter.

4.2. Results and Discussion

4.2.1. n-alkane concentrations of biomass In general the leaves from deciduous trees contain the highest amounts of extractable n-alkanes (as a mean 1066 mg/kg dry weight) but with a high variability between 6540 and 100 mg/kg. No systematic variation of their concentration with species or site was observed. Sphagnum species contain about 91 mg/kg dry weight of n-alkanes (with a variation between 61 to 129 mg/kg). Mosses and Cladonia, only sampled in Scandinavia contain very small amounts of n- alkanes (a mean of 34 mg/kg dry weight with a variation between 8 and 69 mg/kg). Some

31 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient coniferous needle samples from the Scandinavian sites were also extracted, but contained even less extractable n-alkanes than Mosses and Cladonia. The low concentration did not permit the measurement of isotope values, therefore they are excluded from the discussion. These observations suggest that deciduous tree leaves are the major contributors to the sedimentary organic matter of terrestrial origin, with only minor influence by Sphagnum species and a negligible input by Mosses and Cladonia species as well as coniferous tree needles. Also, deciduous trees produce far more biomass than the other plants. However, since sampling inhomogeneity and the extraction procedure are known to influence the measured total concentration of hydrocarbons, we will rely in the following discussion on the relative amounts of n-alkanes. Deciduous tree leaves contain n-alkanes with 25 to 31 carbon atoms and a strong odd over even carbon number predominance (expressed as the CPI) is observed (Table 7), which is in agreement with previous observations (EGLINTON and HAMILTON, 1967; GULZ, 1994;

PIASENTIER et al., 2000). All Betula species in Scandinavia contain also relatively high amounts of n-C23, a substance previously attributed only to water plants (FICKEN et al., 2000) or sphagnum species (BAAS et al., 2000). However, the dominant compound in Betula species is n-C27 in the north Scandinavian samples. In 2 species from southern Scandinavia (LAM) and

Germany (HZM) n-C31 becomes dominant, n-C23 is almost absent from these two samples. The average chain length (ACL) of Betula species is subsequently increasing from N to S from 25.5 (FIN002) to 28.3 (HZM). The ACL index of all analysed species is also increasing from N to S. This suggests that trees in areas with a longer vegetation period and more potential incoming radiation protect their leaves with longer chain n-alkanes from water loss.

The n-C27 alkane is also the dominant compound in Fagus, where it is virtually the only present n-alkane. This is in agreement with previous studies which found high concentrations of n-C27 in beech leaves and soils under beech stands (GLEIXNER et al., 2005; GULZ, 1994).

In all Quercus species, only present in Italy, n-C29 is the major n-alkane, with lower contributions by the n-C25, n-C27 and n-C31 alkanes. The only sampled Alnus species (MAS) contained almost exclusively n-C29, with minor contributions from C27 and C31. One Carpinus species (MEZ) contains n-C29 and n-C31 as the major n-alkanes.

32 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient

Table 7: n-Alkane distributions of the sampled plant biomass: maximum chain length (Cmax), carbon preference index (CPI and CPI23-33) and average chain length (ACL)

sample sampled

code biomass Cmax CPI CPI 23-33 ACL

FIN002 Betula pubescens 27 9.9 13.2 25.5 NAI Betula pendula 27 62.7 89.8 25.6 Moss 27 1.3 4.2 22.3 Sphagnum 27 5.4 14 24.9 KEI Betula pubescens 27 62.0 92.5 25.8 Cladonia 29 0.9 7.1 22.6 SOD003 Betula pubescens 27 34.1 39.5 26.1 Cladonia 27 3.1 3.8 25.6 Fagus sylvatica 27 13.8 14.7 26.8 SOD007 Betula pendula 27 11.5 12 25.8 Cladonia 27 5.0 5.2 25.7 Sphagnum 23 7.1 8.4 25.7 SOD004 Betula pendula 27 12.8 13.7 26.9 Moss 27 1.8 2.8 24.5 HYY Betula pendula 27 16.0 17.2 25.8 Moss 27 7.2 8.6 26.5 Sphagnum 27 10.2 11.5 25.8 SYR Myrte 31 25.3 29.8 30.3 Betula pendula 27 30.9 58.5 26.1 Moss 31 2.3 5.1 26.4 LAM Betula pendula 31 30.2 32.5 28.1 Cladonia 27 3.5 4.3 26.4 HZM Betula pendula 31 15.0 15.5 28.3 Fagus sylvatica 27 59.8 59.8 27 MAS Quercus variabilis 31 2.6 2.8 29.6 Alnus incana 29 6.2 7.1 28.3 Quercus cerris 29 14.4 14.4 28.9 ITA001 Quercus robur 29 15.6 15.6 28.5 MEZ Quercus petraea 29 7.5 8.2 28.3 Carpinus betulus 31 - - 30.3 LGM Fagus sylvatica 27 19.9 26.7 26.9 LPM

One Myrtus plant sampled at SYR contained high amounts of n-C29 and especially n-C31. Also, it was the only sampled plant that contained significant amounts of n-C33. Sphagnum species, only sampled at NAI, SOD007 and HYY, also contain n-alkanes with 23 to

31 carbon atoms, with n-C27 and n-C23 as major compounds. The carbon preference index (CPI) for sphagnum is lower than for deciduous tree leaves, whereas the average chain length (ACL) is in the same range as observed for broadleaf tree leaves (Table 7). Mosses and Cladonia species, sampled in Scandinavia (KEI, NAI, SOD001, SOD007, SOD004, HYY, LAM and SYR) contained not only the long-chain n-alkanes, but also smaller amounts of short-chain alkanes, down to n-C14 in some cases. In the shorter chain n-alkanes, no odd over even carbon number predominance is visible. This leads to much lower CPI values and slightly lower ACL values of these plants compared to Sphagnum and deciduous tree leaves (Table 7).

33 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient

4.2.2. δD values of n-alkanes

The measured δD values on the n-alkanes n-C23 to n-C33 vary between -231 (n-C29 from

Myrtus at SYR) and -141‰ (n-C27 from Quercus at ITA002 and MEZ) (Table 8). Only for the n-C27 alkane of one moss at NAI site a very heavy δD value of -108‰ was observed. Variability of the δD values of the different n-alkanes of one plant sample can be as large as 60‰ (Cladonia at KEI) but averages around 30‰. This variability is larger for Sphagnum, Cladonia and Mosses than for deciduous leaves. In deciduous tree leaves the δD value of n-

C23, if present, is often the heaviest measured δD value.

δD values of the n-C25, n-C27, n-C29 and n-C31 alkanes are lighter in the north and heavier in the south, as expected from the meteoric water δD gradient. Also, the mean δD value (n-C23 to n-

C33) and the concentration weighted mean δD values are in general heavier in the south and lighter in the north (Table 8). n-alkane δD values from Sphagnum species at the measured sites are lower or similar to deciduous tree leave δD values at the same site. n-alkanes from Mosses, which were only sampled in Scandinavia, have higher δD values than Sphagnum or deciduous leaves, expect for SOD007, where δD values are similar to Sphagnum. δD values from Cladonia are similar to δD values from deciduous trees leaves at the same site. δD values of the n-alkanes of different broadleaf trees from the same site have similar δD values. This demonstrates, that the meteorological conditions at a site lead to similar deuterium enrichment in the leaf water. One notable exception is the Myrtus plant sampled at SYR, which is characterized by very low n-alkane δD values of -231, -226 and -225‰ for the n-C29, n-C31 and n-C33 alkanes respectively, about 50‰ lighter than observed for Betula at the same site. Since Myrtus leaves have a very thick wax layer, the leaf is better protected from evaporative water loss, avoiding stronger deuterium enrichment in the leaf water. When comparing the meteoric source water isotopic composition with the n-alkane δD values from the deciduous leaves (Table 9), the best correlation is found for n-C27, the n-alkane, which was found in nearly every plant. Other significant correlations were found for n-C25, n-C31 and the mean and concentration weighted mean δD values of all n-alkanes from n-C23 to n-C31

(Figure 7). n-C23 and surprisingly n-C29, the most abundant n-alkane in the lake sediments analysed for the previous study (SACHSE et al., 2004b) do not show a significant relationship with the source water δD value. However, n-C23 was only found in significant amounts in the

Scandinavian plant samples, whereas n-C29 was most abundant in the Southern European samples. Therefore these 2 alkanes do not cover the whole climatic gradient, which could be the reason for the poor relationship with the source water δD value. The other n-alkanes correlate

34 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient with mean annual δD values, August and September δD values as well as with lake or stream water δD values (except for n-C25) (Table 9, Figure 7).

Figure 7: Correlation of august precipitation δD values and weighted mean n-alkane δD values from deciduous leaves throughout the climatic gradient. The dotted line shows the 95% confidence interval. The error bars are representative of the variability of the concentration weighted mean δD value of the leaf wax n-alkanes within one sample.

None of the correlations is a 1:1 relationship. All equations have a slope lower than one, suggesting that the apparent isotopic fractionation (ε) is lower for lower δD source water values. However, since the isotopic composition of the source water for plants, the meteoric water, is modified in the leaf water due to evapotranspiration, and this enrichment was in fact higher in Scandinavia during Summer 2002, the deviation from a slope of one is a measure of the influence of leaf water enrichment in plants. The highest correlation coefficients were found for the August meteoric water value. Since our sampling took place in August and September, this suggests, that the deciduous tree leaves record the meteoric water from the previous weeks. These data indicate, that n-alkanes are synthesised year-round in the wax layer of leaves. However, the correlation coefficients are only slightly higher than for mean annual or September values, therefore this interpretation remains speculative. Research should focus on possible seasonal variations of n-alkane δD values in the leaf waxes of deciduous trees.

35

Table 8: δD values and standard deviation (SD, 2σ) of the n-alkanes n-C23 to n-C33 and the concentration weighted mean δD values of the sampled biomass

sample sampled weighted

code biomass n-C23 SD n-C24 SD n-C25 SD n-C26 SD n-C27 SD n-C28 SD n-C29 SD n-C30 SD n-C31 SD n-C32 SD n-C33 SD mean SD mean C o m

FIN002 Betula pubescens -168 1 -165 7 -198 1 -170 7 -189 0 -185 15 -187 p o u

NAI Betula pendula -156 7 -187 3 -191 5 -178 19 -182 n d

Moss -108 4 -108 - -108 - s p

Sphagnum -162 2 -197 2 -213 1 -191 26 -197 e c i f

KEI Betula pubescens -157 6 -170 6 -193 2 -173 18 -180 i c

Cladonia -143 6 -167 20 -207 8 -199 5 -179 30 -189 δ D

SOD003 Betula pubescens -171 2 -188 2 -188 0 -183 10 -185 v a l Cladonia -173 1 -173 - -173 u e s

Fagus sylvatica -180 2 -180 - -180 o f

SOD007 Betula pendula -135 15 -162 6 -170 6 -185 10 -193 2 -169 23 -162 n - a

Cladonia -189 1 -201 2 -214 1 -195 0 -204 0 -171 3 -190 1 -179 1 -195 14 -200 l k a

Sphagnum -206 5 -194 3 -191 12 -217 4 -201 13 -202 10 -201 n e s

SOD004 Betula pendula -174 2 -188 1 -176 3 -189 0 -167 0 -166 2 -177 11 -182 f r o

Moss -145 3 -169 10 -176 2 -163 16 -164 m

t HYY Betula pendula -192 1 -210 0 -193 3 -205 0 -189 4 -186 6 -196 11 -203 e r r Moss -175 4 -183 7 -179 5 -181 e s t r

Sphagnum -178 1 -217 5 -200 4 -198 19 -200 i a l

SYR Myrte -231 3 -199 7 -226 2 -210 12 -225 1 -227 4 -228 p l a

Betula pendula -155 4 -183 2 -189 2 -170 15 -182 n t s

Moss -182 6 -198 12 -186 4 -189 9 -190 a l o

LAM Betula pendula -189 4 -183 2 -160 7 -162 2 -174 15 -175 n g

Cladonia -179 8 -179 - -179 a

c l i

HZM Betula pendula -186 3 -175 1 -169 1 -168 1 -174 8 -174 m

Fagus sylvatica -173 1 -173 - -173 a t i c

MAS Quercus variabilis -181 10 -161 3 -171 15 -171 g r a

Alnus incana -174 8 -174 - -174 d i Quercus cerris -159 3 -174 5 -148 3 -161 13 -166 e n t ITA001 Quercus robur -149 4 -141 4 -157 3 -136 6 -146 9 -145 MEZ Quercus petraea -141 6 -161 3 -151 14 -156 Carpinus betulus -153 9 -142 9 -148 8 -146 LGM Fagus sylvatica -149 1 -142 1 -149 3 -147 4 -143 36 LPM

36 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient

Table 9: Correlation of the n-alkane δD values with meteoric water (r is the Pearson correlation coefficient, p – value >0.05 indicates a significant relationship, a and b are the coefficients for the linear regression equation y = ax+b), bold values indicate significant correlations concentration

n-C23 n-C25 n-C27 n-C29 n-C31 mean C23-31 weighted mean C23-31

mean annual δD (OPIC) r -0.265 0.631 0.825 0.174 0.773 0.674 0.67 p 0.526 0.028 0 0.609 0.009 0.001 0.001 a 0.4 0.6 0.5 0.3 0.3 b -146.3 -129.3 -128.7 -147.2 -145.8

lake water δD r -0.428 0.557 0.815 0.098 0.784 0.681 0.66 p 0.29 0.075 0 0.789 0.012 0.001 0.002 a 0.4 0.3 0.2 0.2 b -149.9 -150 -159.3 -159.9

august δD (OIPC) r -0.294 0.677 0.843 0.199 0.772 0.692 0.688 p 0.479 0.016 0 0.557 0.009 0.001 0.001 a 0.5 0.6 0.5 0.3 0.4 b -149.3 -136.1 -136.5 -150.8 -150

september δD (OIPC) r -0.3 0.59 0.801 0.185 0.768 0.659 0.654 p 0.471 0.043 0 0.586 0.009 0.002 0.002 a 0.4 0.6 0.5 0.3 0.3 b -151.9 -134.9 -131.4 -149.7 -148.8

4.2.2. Hydrogen isotope fractionation between source water and biomass n- alkanes The mean hydrogen isotope fractionation between source water (august precipitation) and the weighted mean n-alkane δD values (εw/a) of all plants is -118‰, which is slightly higher than found in the sedimentary record. This might be due to higher evaporative enrichment of the leaf water in summer, whereas the sediments will receive most of the leaves in autumn. However, our sampling might also not be representative for the total terrestrial input into the lake sediments. εw/a varies between -86‰ and -166‰, excluding the exceptionally high value of - 22‰ for the Moss at the NAI site (Figure 8). Accounting for the net biochemical fractionation of hydrogen during biosynthesis of -157‰ (SACHSE et al., 2004b; SESSIONS et al., 1999) this results in an enrichment relative to plant source water of up to 76‰. This agrees well with the range observed in leaf water (ZIEGLER, 1989), which is the source water for biosynthesis. The mean εw/a for all deciduous trees is -121‰, excluding the light Myrtus δD values. For Myrtus at

SYR an εw/a value of -166‰ is observed, suggesting that no or only a very small enrichment of leaf water has taken place. The thick wax layer might be able to restrict evaporative enrichment of leaf water to a minimum. In general, deciduous tree leaves have very similar εw/a values at the same sites (Figure 8), but can vary substantially even between neighbouring sites, as evident for the three finish sites close to the town of Sodankylää (SOD003, SOD004 and SOD007). We conclude, that εw/a in broadleaf tree leaves is a measure of the meteorological conditions at the site. Overall, the plants without stomata, Moss and Cladonia, have smaller mean εw/a values of -

37 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient

87 and -113‰ respectively, and therefore show a stronger isotopic enrichment of leaf water compared to broadleaf trees, which contain stomata. Stomata therefore regulate water loss of higher plants and prevent strong loss of water by evapotranspiration. A small εw/a value of only -22‰ is observed for the moss at the KEI site. This indicates that an extremely strong evaporative enrichment has taken place in the plants tissue, which is reasonable, since mosses can survive dry periods and regenerate, when water is again available.

Figure 8: Comparison of εa/w values for sedimentary n-alkanes (open squares) and deciduous tree leaf n- alkanes (black squares) and non-stomata containing plants (grey squares) for the analysed sites.

Sphagnum species, which also have no stomata, have a mean εw/a of -128‰, which is lower than εw/a values for deciduous tree leaves at the same site. This is reasonable, since Sphagnum grows in a wet environment, evaporative enrichment should not be as strong as for deciduous leaves.

εw/a is higher in northern Scandinavia, lower in central Europe and higher in Southern Italy (Figure 8). Since summer 2002 was exceptionally hot and dry in Scandinavia, but cool and wet in Central Europe, the observed pattern agrees with meteorological observations. Moreover, the average chain length of all analysed biomass samples shows a linear relationship with εw/a (Figure 9). This relationship is equally strong when only considering the stomata containing broadleaf species (Figure 9). With increasing chain length the εw/a value increases, therefore the evaporative enrichment of deuterium in the leaf water decreases. This indicates that longer chain compounds provide a better protection from evaporative loss of water for plants. It has been

38 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient shown that the composition in the leaves of Rosmarinus officinalis changes with changes in moisture and temperature on a seasonal scale (MAFFEI et al., 1993). Therefore, plants might increase the production of longer chain n-alkanes in the leaf waxes as a response to dryer conditions and to prevent evaporative water loss.

Figure 9: Average chain length (ACL) versus εw/a of non-stomata biomass (grey circles), deciduous tree leaf (black circles) and sediment (open circles) samples. The grey line is the linear fit for all biomass samples, the black line is for deciduous tree leaves only. The dotted line represents the 95% confidence interval (only shown for all plants).

However, the observed linear relationship between ACL and εw/a is not observed in the associated lake sediments. The ACL values (n-C25 to n-C31) of the sedimentary n-alkanes vary only slightly along the transect (27.7 to 28.6). The sedimentary record of a lake integrates over several species and receives most if it’s input in autumn, although not exclusively, since strong precipitation events also deliver leaves into the lake. The sedimentary values fall within the mean values of the broadleaf tree leaves (Figure 9), which are the major contributors, and hence represent spatial and temporal average values.

4.2.3. Comparison with sedimentary n-alkane distributions and δD values The distribution of the terrestrial long-chain n-alkanes in the sediments shows a much smaller variability than for the plant biomass. The ACL indices of all sediments are similar with an average value of 28, whereas the ACL indices of broadleaf tree samples vary between 25 and

39 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient

Table 10: Comparison of the fractionation factors between plant n-alkanes and source water to associated sediment values

sample sampled biomass weighted fractionation fractionation

code biomass mean δD C23-33 εw/a biomass εw/a sediment

FIN002 Betula pubescens -187 -108 - NAI Betula pendula -182 -103 Moss -108 -22 Sphagnum -197 -120 -113 KEI Betula pubescens -180 -102 Cladonia -189 -112 -132 SOD003 Betula pubescens -185 -112 Cladonia -173 -99 Fagus sylvatica -180 -106 -121 SOD007 Betula pendula -162 -86 Cladonia -200 -127 Sphagnum -201 -128 -133 SOD004 Betula pendula -182 -108 Moss -164 -88 - HYY Betula pendula -203 -138 Moss -181 -114 Sphagnum -200 -135 -124 SYR Myrtus -228 -166 Betula pendula -182 -116 Moss -190 -125 -124 LAM Betula pendula -175 -109 Cladonia -179 -113 -128 HZM Betula pendula -174 -138 Fagus sylvatica -173 -137 -138 MAS Quercus variabilis -171 -149 Alnus incana -174 -153 Quercus cerris -166 -144 -107 ITA001 Quercus robur -145 -125 - MEZ Quercus petraea -156 -132 Carpinus betulus -146 -121 -121 LGM Fagus sylvatica -143 -118 -142 LPM -127

30 (Figure 9). At the investigated sites it is not possible to determine the dominant broadleaf tree species from the distribution of n-alkanes in the sediments. The observed n-alkane distributions, for instance for Betula, show, that even the same species can have variable n-alkane distributions. Also, seasonal variations in n-alkane distributions in leaves have been observed

(GULZ, 1994). The vegetation at site HZM consist of almost exclusively of Fagus species, which only contain the n-C27 n-alkane. However, the sedimentary record of lake HZM shows a

40 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient

uniform distribution of the n-C25 to n-C31 n-alkanes. Therefore, the use of n-alkane distributions in lake sediments as a chemotaxonomic indicator is not recommended.

The mean δD values of the sedimentary n-C25 to n-C31 alkanes (n-C23 from the sediments is omitted, since its lower δD values suggest they originate from aquatic plants rather than terrestrial plants) taken from the previous study (SACHSE et al., 2004b) are heavier than the weighted mean plant biomass δD values at nearly all sites (Figure 10, Table 10).

Figure 10: Comparison of the concentration weighted mean broadleaf tree n-alkane δD values (filled circles), Sphagnum δD values (open circles) with mean sedimentary n-alkane δD values.

The n-alkanes from Myrtus at SYR have 20‰ lighter δD values than the sediment, possibly due to their thick protective wax layer as discussed above. The leaf wax n-alkanes from MAS have 40‰ lighter δD values. Since all 3 sampled species from MAS have virtually the same δD and εw/a values and the big size of the lake and its catchment area compared to the other lakes, these values might not be representative of the lake catchment. All other plant biomass δD values are heavier, some values equal, than the associated sediment n-alkane δD values (Figure 10). Since our sampling took place in late summer, when evapotranspiration and therefore leaf water isotopic enrichment should be high, this observation is reasonable. In autumn, when most leaves will fall into the lakes and get incorporated into the sedimentary record, a lighter δD value would be expected, because of lower evaporation of leaf water. Consequently, the plant n- alkane δD values seem to record seasonal variations.

41 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient

Sediment as well as plant εw/a values show the same pattern: higher values in Northern Scandinavia, lower values in Central Europe and higher values in Italy (Figure 8), suggesting that the trend of leaf water enrichment is preserved in the sedimentary record. These findings indicate that changing δD values of deciduous tree derived n-alkanes in the sedimentary record can be used as a proxy for changes in leaf water enrichment and therefore environmental factors such as relative humidity and precipitation.

4.3. Conclusions The comparison of n-alkane δD values and concentrations from different plants along a climatic gradient and associated sedimentary n-alkane δD values indicates that:

- the relative abundance of specific n-alkanes within one species is not constant over the transect, longer chain n-alkanes are present in Betula species from Southern Scandinavia and Central Europe compared to the Northern Scandinavia. This makes the use of n- alkane distributions in sediments as an indicator for specific species nearly impossible.

The n-C23 alkane, previously only attributed to submerged aquatic plants and Sphagnum was also found in significant amounts in Betula leaves. Therefore care should be exercised when using sedimentary n-alkanes to separate aquatic and terrestrial organic

matter input, since n-C23 can be derived from both sources. These findings explain the

high variability of n-C23 δD values observed in lake sediments in the previous study (SACHSE et al., 2004b).

- δD values of different deciduous tree leave n-alkanes are similar within a site. They are variable even within species, when comparing different, neighbouring sites. However, different water use strategies are visible between deciduous trees and Sphagnum species, which are characterized by lighter δD values (and therefore less evaporative enrichment of water in the tissue) at the same site, since they grow in wet environments. Mosses and Cladonia, which contain no stomata to regulate water loss, have similar or heavier δD values than broadleaf trees at the same site. These results suggest that plant anatomy (stomata vs. non-stomata containing plants) and site conditions on a small scale (relative humidity, soil moisture, precipitation) and not biosynthetic differences among

species are the major drivers for variations in εw/a.

42 Compound-specific δD values of n-alkanes from terrestrial plants along a climatic gradient

- δD values of the deciduous tree leave n-alkanes correlate highest to the august precipitation δD values. This indicates that seasonal variations in n-alkane δD values do exist. Also, sedimentary n-alkane δD values are in general similar or lower than the sampled plant material from the summer. Further research on the timescale and magnitude of seasonal variations in leaf wax n-alkanes is necessary to fully understand the origin of the terrestrial δD signal in the sedimentary record.

- the average chain length (ACL) of the dominating vegetation is increasing from N to S, suggesting that plants subject to higher degrees of evaporation protect their leaves with longer chain compounds. Evidently, a significant negative relationship is observed between the ACL value of the biomass and the isotopic fractionation between source

water and n-alkanes (εw/a,) suggesting that longer chain compounds indeed provide a better protection from evaporative water loss through the leaves.

These results prove, that leaf wax n-alkane δD values from terrestrial plants, especially of broadleaf trees, record the δD value of the precipitation, modified by site specific meteorological conditions (evapotranspiration, relative humidity and soil moisture). The comparison of biomass and sedimentary n-alkane distributions and δD values shows that the sedimentary record integrates spatially as well as temporary. Therefore it is suggested that variations in the terrestrial biomarker δD values in lake sediments can be used as an indicator of changing climatic conditions, if the water source is not changing.

43 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

5. Seasonal variations in deciduous tree leaf wax n-alkane δD and δ 13C values: Implications for their use as a palaeoclimate proxy

5.1. Introduction The stable isotope ratios of hydrogen (δD values) and oxygen (δ18O values) in meteoric water depend on a variety of climatic parameters, such as temperature, humidity and evaporation

(CRAIG, 1961; GAT, 1996). Since meteoric water is the major oxygen and hydrogen source for nearly every organism on Earth, the potential of organic matter to record the meteoric water isotope signal has been a focus of research. Cellulose was found to record the δD and δ18O value of the meteoric water and has been applied in palaeoclimate reconstruction using tree rings

(FENG and EPSTEIN, 1995; YAPP and EPSTEIN, 1982). With growing interest in this application, the associated fractionation processes have been studied in detail. It is well known that the water in the leaves of plants becomes enriched in the heavy isotopes 18O and D during evapotranspiration processes (LEANEY et al., 1985). Models have been developed to describe the isotopic enrichments of leaf water compared to plant source water (CRAIG and GORDON,

1965; FLANAGAN and EHLERINGER, 1991; RODEN and EHLERINGER, 1999). These models have shown, that the isotopic enrichment in leaf water is mainly controlled by changes in relative humidity, but also rate (BARBOUR and FARQUHAR, 2000). A number of studies have successfully reconstructed changes in the hydrological cycle using tree-ring cellulose δD 18 and/or δ O values (MAYR et al., 2003; MCCARROLL and LOADER, 2004; PENDALL et al., 1999). Recently, the compound-specific hydrogen isotope composition of lipid biomarkers in sediments has been demonstrated to record the source water δD value with high fidelity (HUANG et al., 2004; SACHSE et al., 2004b; SAUER et al., 2001; SESSIONS et al., 1999). The biosynthetic fractionation of the hydrogen isotopes during biosynthesis was found to be independent of environmental parameters; for n-alkanes an isotopic fractionation between source water and n- alkanes of -157‰ was observed (SACHSE et al., 2004b; SESSIONS et al., 1999). Lipids like n- alkanes are much more stable and more abundant in the sedimentary record than cellulose, making them ideal candidates for reconstruction of changes in the water cycle over geological timescales. Furthermore, n-alkanes contain only carbon-bound hydrogen, which is non- exchangeable (SCHIMMELMANN et al., 1999), and are relatively easy to extract and purify. These compounds hold information on their biological sources, since n-alkanes of different chain- lengths are produced by different organisms. Algae, for example, mainly produce n-alkanes

44 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

with 17 and/or 19 carbon atoms (n-C17, n-C19), submerged aquatic plants are dominated by n-

C23 (FICKEN et al., 2000) and the long-chain n-alkanes with odd carbon numbers n-C25 to n-C31 are found in the leaf waxes of higher terrestrial plants (EGLINTON and HAMILTON, 1967). Large amounts of these alkanes are found in the leaves of broadleaf trees, whereas coniferous tree needles contain much smaller amounts. The different biological sources of n-alkanes permit a differentiation of aquatic and terrestrial organic matter found in lake sediments. Sachse et al. (2004b) examined recent sediments of small ground-water fed lakes along a climatic gradient from northern Finland to Southern Italy and found that the terrestrial derived n-alkanes are enriched in D relative to the n-alkanes of aquatic origin by 10 to 60‰. Sachse et al. (2004b) suggested that if the plants use the same meteoric water source as the aquatic organisms, the isotopic difference between the terrestrial and aquatic n-alkanes in lake sediments would be a proxy of leaf water enrichment and consequently the amount of evapotranspiration and relative humidity. However, it has not been demonstrated up to now a) if n-alkanes show seasonal variations in δD values b) if leaf wax n-alkane δD values record the source water δD signal as a cumulative annual signal or c) if leaf wax n-alkanes δD values record variations in the source water over a certain timeframe In this study those issues will be addressed in order to decipher the origin of the δD signal from terrestrial biomarkers in the sedimentary record. Leaf wax n-alkane concentrations and δD and δ13C values obtained from 2 sites, a Maple and a Beech stand in Germany, are compared over the course of the 2004 growing season. Additionally, the relationship between δD values from the Beech site and meteorological data, monitored throughout the year at the site is explored. The measured δD values from the Beech leaf n-alkanes are compared to modelled leaf water isotope values.

5.2. Materials and Methods

5.2.1. Study site and sample collection Both sampling sites are shown on Figure 1. The Maple site is located in south-western Germany, close to the village of Thann, about 45 km NE of Regensburg. The forest area (2 ha) consists of Maple (Acer pseudoplatanus), planted under an umbrella of Pine and Spruce (Pinus sylvestris and Picea abies). Shaded Maple leaves were sampled every one to two weeks along a path into the forest.

45 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

The Beech sampling site is located within the “Hainich National Park”, near the city of Eisenach in Central Germany (51°04’46”N, 10°27’08”E, 440 m a.s.l.) in a suboceanic/subcontinental climate with an annual mean air temperature of 8°C and 750-800 mm average annual precipitation. The area has been protected as a National Park since 1997, since it is one of the largest virgin Beech forests in Europe. The forest is dominated by Beech (Fagus sylvatica with 65%), Ash (Fraxinus excelsior, 25%) and Maple (Acer pseudoplatanus and plantonoides, 7%). Other deciduous and, to a smaller extent coniferous trees are also present. The lack of management has resulted in a wide distribution of tree age classes with a maximum of up to 250 years. A more detailed site description is given in Knohl et al. (2003). Sampling of leaves took place approximately every two weeks during the 2004 growing season (April to early November) along a ca. 50 m transect from a track to the meteorological measurement tower site. For both sites approximately 20 shaded leaves from different trees at a height of about 2 m were pooled for analysis, to obtain a representative, average value..

5.2.2. Meteorological measurements at the Hainich site Meteorological and ecosystem data were measured at an eddy covariance flux tower, whose footprint covers the sampled area at the Hainich site. Air pressure (PTB101B, Vaisala, Helsinki, Finland), air temperature (at 2 m) and air humidity (HMP35D, Vaisala, Helsinki, Finland), photosynthetically active radiation (LI-190SA, LiCor Inc., Lincoln, NE, U.S.A.) and water vapour concentration (LiCor 6262-3, LiCor Inc., Lincoln, NE, U.S.A.) were monitored continuously. Precipitation was collected at the tower site (RainGauge, Young, Traverse City, MI, U.S.A.). In addition, soil moisture was measured at 8 cm depth (Theta probes, ML-2x, Delta T, Camebridge, U.K.). A detailed description of the available measurement equipment at the Hainich site is given in Knohl et al. (2003).

5.2.3. Modelling leaf water isotopic enrichment The classical model describing the enrichment of 18O and D of a water body due to evaporation was developed by Craig & Gordon (1965). The model was later expanded to include leaf boundary layers (FARQUHAR et al., 1989; FLANAGAN and EHLERINGER, 1991). To calculate the leaf water isotopic composition the model described in (RODEN et al., 2000; RODEN and

EHLERINGER, 1999) will be used. It should be noted that this model predicts the isotope composition of leaf water at the site of evaporation, e.g. the , where the and carbohydrate metabolism take place. In contrast, bulk leaf water in plants is

46 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

isotopically more heterogeneous (YAKIR et al., 1989), due to specific anatomical, morphological and physiological features of leaves. Since the isotope composition of the hydrogen source for biosynthesis will be modelled, there is no need to account for these processes. The model is available as an electronic spreadsheet for download under ftp://ecophys.biology.utah.edu/Tree_Ring/:

*   ei − es   es − ea   ea   Rwl = α α kRwx   + α kbRwx   + Ra    Equation 2   ei   ei   ei  

The subscripts wl, wx and a refer to leaf water, xylem water (which is equivalent to the source water for the plant) and bulk air, respectively. e is the vapour pressure (subscripts i, s and a refer to intercellular air spaces, leaf surface and bulk air, respectively). The fractionation factors involved are well established: the equilibrium fractionation between liquid and water vapour (α*) is slightly temperature dependent (MAJOUBE, 1971), the kinetic

fractionation associated with diffusion in air (αk) has been determined by Merlivat (1978). The

kinetic fractionation during diffusion through the boundary layer (αkb) is calculated raising αk to the 2/3 power (FLANAGAN et al., 1991).

The major environmental factors controlling the hydrogen isotope composition of leaf water (Rwl

or δDwl) are the relative humidity (which is used in the calculation of ea) and the isotopic

composition of the source water (Rwx) and the atmospheric water vapour (Ra). ei is a function of

leaf temperature. The vapour pressure at the leaf surface, es is the only parameter where leaf physiology is considered, and it is estimated using stomatal conductance and transpiration rate. However, only differences in an order of magnitude in the estimation of stomatal conductance and transpiration rate are likely to affect the modelled leaf water isotopic composition by a few

‰ (RODEN et al., 2000). As model input a value of 0.1 mol m-2 s-1 for stomatal conductance is used, which is reasonable

for Beech leaves (HERBST, 1995) and a value of 1 mol m-2 s-1 for the boundary layer

conductance, which is used estimating leaf transpiration and finally es. Since the temperature at the leaf surface, where the modelled processes take place is rarely equal to air temperature, α* is calculated using a 2° higher leaf temperature, which is true for high radiation loads, when most photosynthetic activity is expected. However, high transpiration rates and low radiation can result in a cooler leaf than air temperature. It should be noted, that the sensitivity of the model to changes in leaf temperature is rather small, a 1°C decrease in temperature results in about 1‰ decrease in leaf water δD values, which is within the standard

deviation of our measurement. αk is assumed to have a value of 1.025 (MERLIVAT, 1978).

47 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

For δi, the source water isotopic composition, the monthly δD values of meteoric water, calculated using the Online Isotopes In Precipitation Calculator (OIPC, accessible at http://www.waterisotopes.org), which uses the IAEA-GNIP meteoric water database (IAEA, 2001) and interpolation algorithms described in Bowen & Revenaugh (2003) are applied. The calculated isotopic composition of meteoric water was found to accurately predict the actual measured isotopic composition of lake and stream waters throughout Europe (SACHSE et al., 2004b).

The isotopic composition of the atmospheric water (δDa) exerts a major influence on δlw. Since there are no direct measurements of δDa, these values have been estimated using empirical, temperature dependent equations obtained by Jacob & Sonntag (1991). The authors investigated an 8-year record of precipitation and water vapour at Heidelberg in Southern Germany, where the general climate is comparable to our study sites. Using the daily mean temperature measured at the tower one obtains a mean δDa value of -132.3‰ (n=228, SD=13.3) for the period between the 1.4. 2004 and the 15.11.2004. The mean δD value of meteoric water obtained using the OIPC for this period is -56.0‰ (using the equations by Jacob & Sonntag (1991) for estimation of δD of the precipitation, results in a rather similar mean value of -55.1‰), therefore the mean isotopic difference between water vapour and precipitation is -76.2‰. Under equilibrium conditions at 20°C a mean isotope fractionation of -85.1‰ would be expected, which is similar to the empirically estimated values, when considering the mean standard deviation of the n-alkane δD measurement of 4‰. Therefore a steady state between water vapour and precipitation is assumed, which is true only for the wetter periods. However, using the above mentioned estimations results in reasonable values. Therefore, the mean isotopic difference between atmospheric water vapour and precipitation at our site is assumed to be -

76.2‰ and δDa values are calculated accordingly.

5.3. Results

5.3.1. n-alkane concentrations 1. Thann site (Maple) The total concentration of n-alkanes from Maple leaves ranges between 25 and 100 mg per kg dry weight. The total concentration of n-alkanes is highest in spring and fairly constant over summer, before decreasing in August. However, we will rely here more on the relative amounts of n-alkanes, since sampling heterogeneity and the extraction procedure might be responsible for small concentration changes. Maple leaves contain the n-alkanes n-C25, n-C27, n-C29 and n-

C31. n-C25 makes up 6% of the annual mean and is only a minor constituent. The major n-alkane

48 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

is n-C29, with 38%, whereas n-C27 and n-C31 both compromise about 28%. The amount of n-

C25 is rather constant over the growing season, whereas the relative amount of n-C27 and n-C31 is decreasing from 38% in May to 22% and from 37 to 17% in November, respectively. n-C29 is constantly increasing from 25 to 57% over the course of the sampling period (Figure 11).

Figure 11: Seasonal variations in the relative amounts of n-alkanes from Maple (top) and Beech (bottom) leaves

Hainich site (Beech) The total concentration of n-alkanes from Beech leaves ranges between 150 to 260 mg per kg sample dry weight, and is slightly higher than for Maple leaves. The highest total concentrations were also observed in spring, with decreasing amounts over the summer peroid. The analysed

Beech leaves contained only 6% of n-C25 and 4% of n-C29 as annual means. The major constituent was n-C27, which made up 90% of all n-alkanes in Beech leaves. No change in the relative amount of n-alkanes over the growing season was apparent (Figure 11).

49 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

5.3.2. n-alkane δD and δ 13C values

1. Thann site (Maple)

Isotopic ratios were measured on the n-alkanes n-C27, n-C29 and n-C31 extracted from the Maple 13 leaves. The δ C values of n-C29 decreased throughout the growing season from -37.1 to - 13 38.4‰, whereas the δ C values of n-C27 and n-C31 were about 2‰ heavier and increased throughout the year by about 1‰ (Table 11, Figure 12).

13 Table 11: δ C and δD values of the n-C27 n-C29 and n-C31 alkanes extracted from Maple leaves during the 2004 growing season

Thann (maple) n-C27 n-C29 n-C31 sampling date δ13C stdv δD stdv δ13C stdv δD stdv δ13C stdv δD stdv

03.05.2004 -37.2 0.1 -154 1 -37.1 0.1 -169 6 -36.5 0.1 -178 1 23.05.2004 -38.4 0.3 -158 3 -38.2 0.1 -159 19 -37.4 0.6 -169 4 18.06.2004 -37.1 0.1 -169 0 -38.7 0.3 -181 4 -37.3 0.5 -178 5 22.07.2004 -36.8 0.2 -158 3 -38.3 0.3 -183 1 -36.4 0.0 -179 3 28.08.2004 -37.1 0.1 -142 8 -38.3 0.1 -175 7 -36.5 0.3 -164 7 09.09.2004 -37.5 0.3 -129 10 -38.2 0.2 -169 6 -37.1 0.2 -157 9 12.10.2004 -37.2 0.1 -172 2 -39.2 0.2 -187 5 -36.8 0.0 -177 2 20.10.2004 -37.0 0.8 -137 1 -39.0 0.4 -176 2 -37.1 0.3 -159 7 08.11.2004 -36.6 0.2 -178 1 -38.4 0.2 -192 3 -35.2 0.2 -182 3

δD values of all 3 n-alkanes were highly variable. n-C29 was characterized by the lowest δD values, n-C31 was about 5-10 ‰ heavier and n-C27 was again 10-20‰ heavier. All three n- alkanes showed a similar variability over the year. Higher values were observed in early May. A subsequent rise in δD values was apparent until late August, the values decreased again in late September, before showing another increase in early October. To mid-October, δD values of all three n-alkanes showed a decrease.

13 Table 12: δ C and δD values of the n-C27 alkane from Beech leaves

Hainich (beech) n-C27 sampling date δ13C stdv δD stdv

30.04.2004 -35.5 0.1 -156 1 17.05.2004 -35.8 0.1 -159 2 26.05.2004 -34.8 0.1 -168 3 11.06.2004 -36.9 0.1 -175 3 24.06.2004 -36.0 0.1 -158 3 07.07.2004 -35.4 0.3 -168 1 06.09.2004 -33.6 0.1 -160 4 15.09.2004 -34.3 0.1 -160 7 29.09.2004 -35.2 1.1 -140 4 14.10.2004 -34.8 0.2 -159 3 27.10.2004 -34.5 0.6 -156 2

50 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

2. Hainich site (Beech)

13 δD and δ C values were determined only on the n-C27 alkane, the concentration of the other compounds was too low for the measurement of isotope ratios (Table 12, Figure 12). 13 δ C values of the n-C27 alkane rose to a value of -34.8‰ in late May, before they dropped to - 36.9‰ in June. A high value of -33.6‰ was observed again in early September, before values stabilized at around -35‰.

δD values of the n-C27 alkane decreased from a relatively high value of -155‰ in late April to - 180‰ in early June. An increase in δD values was observed in late June. Another strong increase with the highest measured value in the 2004 growing season, -140‰ was observed in early October. Afterwards values decreased to -160‰. It is evident, that mean δD values in September and October were slightly higher than from April to July.

Figure 12: Seasonal variations in the δD and δ13C values of n-alkanes from Maple (top) and Beech (bottom) leaves

51 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

5.3.3. Modelling leaf water δD values at the Hainich site The modelled leaf water δD values for the Hainich site vary between +10‰ and -45‰ (daily values) or 0 to -30‰ as 3-day mean values over the growing season, with a mean value of - 17‰ (Figure 13). This results in an average enrichment over mean annual source water δD value (meteoric water δD = -56‰) of 39‰. The modelled δD leaf water values correlate with relative humidity (r2=0.92, n=260), demonstrating the strong dependence of leaf water δD values on relative humidity. The highest δD values were modelled for late April, late June (up to +8‰) and early September (up to +12‰), coinciding with longer periods of low relative humidity and few rainfall events (Figure 15). Towards the end of the growing season modelled

δDwl values decrease to -40‰ in late October.

5.4. Discussion

5.4.1. Seasonal variations in n-alkane concentration

Maple leaves contained mostly the n-C27, n-C29 and n-C31 alkanes (Figure 11).

The observed seasonal variation, an increase in the relative amount of n-C29 over the growing season and a decrease in the relative amount of n-C27 and n-C31 suggests that n-alkanes are synthesised de novo over the course of the year in Maple leaves.

No change in the relative amount of n-alkanes in Beech leaves was observed, with n-C27 making up about 90% of all n-alkanes. This supports previous observations, where Beech leaves and soils under Beech stands were found to contain high amounts of n-C27 (GLEIXNER et al., 2005; PIASENTIER et al., 2000). However, Piasentier et al. (2000) reported a progressive increase in the longer-chain n-alkanes during successive leaf stages from to mature leaves for Ash, Hazel and Beech. This phenomenon is observed in the Maple leaves with the increasing relative amount of n-C29, but not for Beech leaves. Piasentier et al. (2000) also report higher amounts of n-C25 in young Beech leaves than found in this study. This suggests that intra-species variability and/or environmental factors play a role in n-alkane synthesis in the leaf wax layer as observed for Rosmarinus officinalis (MAFFEI et al., 1993).

5.4.2. Seasonal variations in n-alkane δD and δ 13C values for Maple leaves δD and δ13C values of leaf wax n-alkanes of two deciduous trees are highly variable throughout the year (Figure 12). Those results again suggest, that n-alkanes in the leaf wax layer of higher plants are synthesised de novo year-round. The observed maximum variations of the n-alkane δD values of about 40‰ were within the range of expected leaf water enrichment (LEANEY et

52 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values al., 1985). However, the question remains, if leaf wax n-alkanes preserve a cumulative leaf water isotope signal, or if complete or partial turnover of the leaf wax n-alkanes results in an integrated value over a certain timeframe. According to Gulz (1994) the leaf wax is completely developed shortly after unfolding of the leaf in spring but several studies have shown that the chemical composition of the wax layer is constantly changing (GULZ, 1994; PIASENTIER et al., 2000). This would imply that the deuterium content of the n-alkanes should vary accordingly. The decreasing total n-alkane concentration over the growing season suggests that more n- alkanes are produced in spring than in summer. However, the changing relative amounts of n- alkanes for the Maple leaves and the high variability of the n-alkane δD values, especially in summer, for Maple and Beech leaves suggest de novo production all over the year. It is known that abrasion by wind removes significant amounts of n-alkanes from the leaf wax layer

(SCHEFUSS et al., 2003), which would imply the need for a year-round production of n-alkanes, to ensure the protection from evaporative water loss from the leaves. The δD values of n-alkanes from the Maple leaves from Thann show relatively high values at the beginning of the 2004 growing season in late April and early May, before they decrease to lower values in June. Although this might be related to an enriched source water, it is more likely that isotopically enriched reserves, mainly stored as from the previous year, are used up during the for the biosynthesis during the early growth season. It is known that tree growth in early spring mainly depends on reserves from the previous year (DAMESIN and

LELARGE, 2003), which has also been observed in tree-ring cellulose from early- (HELLE and SCHLESER, 2004). Thus, the δD signal preserved in the leaf wax n-alkanes in spring might reflect dry conditions from last autumn and not the current spring conditions. The relatively high difference between δD values of the different n-alkanes (as much as 50‰ for the August event between n-C29 and n-C27, Figure 12) suggests a pronounced heterogeneity in the hydrogen source. Either the more enriched substances (n-C27 and n-C31) are preferably synthesised during these two events (although no significant change in the relative amount of these n-alkanes is observed at that time), or they are synthesised from a more enriched hydrogen pool than n-C29. The δ13C values of the n-alkanes from Maple leaves vary only slightly by 2‰ over the growing season (Figure 12, Table 11). Interpretation of δ13C values from biomass values is difficult, since several factors can influence these values. The discrimination of 13C in biomass during carbon fixation can be expressed as:

53 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

 ci  ∆ = a + (b − a)  Equation 3  ca 

(FARQUHAR et al., 1982)

The net discrimination consists of 2 processes: the discrimination during diffusion of CO2 through the stomata (a), which is estimated to be -4.4‰ and the net discrimination due to carboxylation (b) which is about -27‰. However, the biosynthesis of specific compound classes, which are synthesised from the products of the Calvin Cycle, involves additional discriminations. Lipids, for example, are generally more depleted in 13C than bulk biomass

(ABELSON and HOERING, 1961; COLLISTER et al., 1994).

ci and ca are the intercellular and ambient CO2 concentrations. This ratio is controlled by stomatal conductance and photosynthetic rate. Variation of both parameters is dependant on physiological factors (eg. different species have different responses of stomatal conductance to environmental parameters) and environmental factors. Separation of the relative influences of these different factors is not straightforward. However, Yakir & Israeli (1995), have suggested that with the combined use of δ18O and δ13C values of leaf tissue it is possible to separate these two factors. Scheidegger et al. (2000) have presented a conceptual model to separate the relative influences of stomatal conductance and photosynthetic rate based on δ18O and δ13C values of organic matter. Since δD values of leaf water, the hydrogen source for the biomass, are subject to the same processes as δ18O values, their responses to the discussed processes should be rather similar. The δ13C values obtained from the Maple leaves do not show any positive or negative correlation with δD values, their variation over the growing season is very small 13 18 (within 2‰). Little effect on δ C values during a change in δ O (eg. δD) of biomass would be due to simultaneous changes in photosynthetic rate and stomatal conductance (SCHEIDEGGER et al., 2000). Following this interpretation a high δD values with no apparent changes in δ13C, as observed in early September and October for the Maple leaves, would be caused by reduced photosynthetic rate and reduced stomatal conductance during dryer periods. Low δD values without changes in δ13C, as observed in early October, are caused by the opposite effect during wetter periods.

5.4.3. Seasonal variations in n-alkane δD and δ 13C values for Beech leaves - Comparison of modelled leaf water enrichment with n-alkane δD values

The δD values of n-C27, the only major n-alkane found in the Beech leaves from the Hainich site, also start with relatively high values in late April (Figure 12). This again might not be due

54 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values to leaf water enrichment, but to the use of stored reserves from the previous year. Two enrichment events are observed in n-C27 δD values, the first in June and a second, more pronounced event in late September. If the leaf water δD value is calculated from the δD value of the n-C27 alkane, using a mean biosynthetic fractionation during hydrogen incorporation into n-alkanes of -157‰ (SACHSE et al., 2004b; SESSIONS et al., 1999), values vary between +10‰ and -15‰, well within the range of the modelled δD values (Figure 13), suggesting that the n-

C27 δD value is determined by δD of the leaf water. Since it is not expected that the n-alkane δD value on the sampling day records the current δD leaf water signal, a possible time lag has been investigated. 3-day mean values of modelled δD for subsequent 3, 9, 12, 15, 18, 21, 24, 27 and 30 day time lags were calculated for the two observed enrichment events and compared to the measured n-C27 δD values (the high late April value is left out, since it might reflect the use of storage products rather than leaf water as a hydrogen source). Significant correlations are obtained for a time lag of 27 (r=0.99, n=5) and 30 days (r=0.93, n=5) for event one and of 18 days (r=0.95 n=5) for event 2 (Figure 14). For event one relative humidity shows also a negative relationship with n-alkane δD values (r=-0.75 and r=-0.73 for 27 and 30 days respectively). However, it is not statistically significant. For event two relative humidity gave a significant negative relationship to δD of n-C27 for an 18-day time lag (r=-0.90, n=5). The resulting equations for the linear regressions are shown in Table 13.

Figure 13: Modelled leaf water δD values and estimated δD values from the measurement of the n-C27 alkane for the Hainich (Beech) site

The relationship for the 27-day time lag, with the highest Pearson coefficient (r=0.99) between modelled leaf water δD and measured n-C27 δD is nearly a 1:1 relationship (slope is 1.04) with

55 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values an offset of -161.3‰ between leaf water and n-alkane. This offset is strikingly similar to the expected biosynthetic fractionation between source water and lipid of about -157‰, demonstrating that the n-alkane δD value tracks the leaf water isotopic composition. For event 2 and an 18-day time lag a steeper slope (1.24) with an offset of about -167.9‰ between leaf water and alkane is observed. Figure 13 shows that event 2 is characterised by the highest enrichment in the n-C27 δD values observed during our study. Roughly 18 days before a strong enrichment in leaf water δD was also found in our model runs, however, the predicted enrichment is about 15‰ lower than the observed enrichment in the n-C27 alkane. When taking into account the soil moisture at 8 cm depth at the study site and the amount of precipitation over the 2004 growing season (Figure 15) it becomes apparent, that August, September and October of 2004 were much drier than May, June and July. The late summer/autumn soil moisture content was reduced to 2/3 of the spring/early summer value. According to a simple Rayleigh fractionation model for evaporation of water, a loss of 1/3 of the water results in an enrichment of 35‰ at 20°C.

Table 13: Significant correlations for 3-day mean values of modelled δDwl and measured relative humidity (rH) against δD value of the n-C27 alkane for specific time lags. The Pearson correlation coefficient (r) and the resulting linear regression equation are shown. (Note: for n=5 r>0.88 or r<-0.88 is statistically significant)

δD n-C27 vs: δDwl rH r equation r equation Event1

27 days timelag 0.99 δDwl = 161.3 + 1.04*δDC27 -0.75 rH = -97.5 - 1.02*δDC27

30 days timelag 0.93 δDwl = 174.3 + 1.11*δDC27 -0.73 rH = -140.9 - 1.28*δDC27

Event 2

18 days timelag 0.95 δDwl = 167.9 + 1.24*δDC27 -0.90 rH = -155.0 - 1.53*δDC27

Figure 14: Plot of the Pearson correlation coefficients of modelled leaf water δD versus n-C27 alkane δD (open circles) and relative humidity (filled circles) for the specific time lags (3 day means). Above or below the dotted lines correlations are significant (α=0.05).

56 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

However, a Rayleigh fractionation assumes a humidity of 0%, whereas the actual humidity will be higher and the resulting enrichment lower. This effect explains the higher than expected δD leaf water values, determined from the n-C27 alkane δD value and their magnitude of ca. 20‰. This suggests that deciduous tree leaf wax n-alkane δD values track changes in the leaf water δD value by a 2 to 4 weeks time lag. It is not expected that n-alkanes will record every single excursion in leaf water isotope composition, but integrate over longer time periods. Longer dry or wet seasons (in the order of weeks) seem to be recorded in n-alkane δD values. Our data suggest that a more or less complete metabolic turnover of the hydrogen pool in n-alkanes occurs in roughly 4 weeks under these climatic conditions. Consequently, the n-alkanes would not be synthesised from recent assimilates, but would be conversion products formed from earlier formed substances, which carry an older source water signature. This mechanism would explain the rather long time lag of the leaf water hydrogen isotope signal, as observed in our study. Jetter & Schäffer (2001) have shown that after a complete removal of the wax layer of Prunus laurocerasus leaves a rapid reestablishment occurs. After 5 days a complete new wax layer had developed, with acetates as major constituents. Acetate concentration is decreasing at the same time as increase, which become dominant after 20 days, before concentrations of alcohols decrease. n-Alkanes show a slow but steady increase, becoming the dominant compound in the leaf wax ca. 30 days after the disruption. These findings suggest, that a conversion of acetates to alcohols by acetate hydrolysis takes place in the leaf tissue (JETTER and SCHAFFER, 2001). Although P. laurocerasus is different from Beech in several aspects, the data support our findings that n-alkanes seem to have slower turnover rates and therefore excursions in leaf water δD appear in the n-alkanes only after 2-4 weeks. The δ13C values of the Beech leaf wax n-alkanes show a 3‰ variation over the growing season, which is slightly higher than observed for Maple leaves. Also, variations occur on shorter 13 timescales than in the Maple leaves. In late May δD and δ C values of the n-C27 alkane both fall before rising again in late June. Such a pattern could be the result when stomatal conductance exerts the major control, whereas the photosynthetic rate is not changing. Low δ13C values indicate a lower stomatal conductance and higher δ13C values are caused by higher stomatal conductance. Such a pattern would be expected when water is not limited

(SCHEIDEGGER et al., 2000). The relatively high soil moisture content in June (Figure 15) at the Beech site supports this interpretation. The highest single precipitation event during the 2004 growing season occurred on May 7th with nearly 50 mm of rainfall on one day.

57 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

Figure 15: Relative humidity (rH), soil moisture at 8cm and precipitation amount for the 2004 growing season at the Hainich (Beech) site. Strong lines indicate 3-day mean values for rH and soil moisture.

This led to a significant increase in soil moisture in the following days (Figure 15). Due to sufficient water availability stomata will open, leading to a higher stomatal conductance. Higher stomatal conductance results in higher δ13C values of biomass, when photosynthetic activity is 13 constant (SCHEIDEGGER et al., 2000). Interestingly, a peak in the n-C27 δ C value is observed 3 weeks after the precipitation event on 26th of May. Evidence for a rather constant photosynthetic activity is provided by the constant ratio of vapour pressure deficit (VPD) to photosynthetic activity (PAR) in May and July (Figure 16). During September and October lower δ13C values are observed at the same time as high δD values are measured. This isotope pattern suggests that stomatal conductance stays rather constant, but that the photosynthetic rate is lower

(SCHEIDEGGER et al., 2000). To reduce water loss, stomata stay closed during the observed dryer period in autumn, so photosynthetic activity determines changes in biomass δ13C values. The VPD/PAR ratio shows much higher fluctuations from July to October (Figure 16), with lower values in late August caused by low VPD and higher values in early September, indicating less favourable conditions for photosynthesis. 58 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

13 Figure 16: δ C values of the n-C27 alkane and the VPD/PAR ratio (vapour pressure deficit normalized to photosynthetically active radiation) over the 2004 growing season at the Hainich Beech site. The grey line indicates 3-day mean values for the VPD/PAR ratio.

Two weeks after the high VPD/PAR values in early September low δ13C values are observed in the n-C27 alkane, which is in agreement with the interpretation of lower photosynthetic rate as the driver for the observed lower δ13C values. The observed variation in the n-alkane δ13C values can be explained in conjunction with their δD values. The observed 2-4 week time-lag for δD in n-alkanes is also evident in δ13C values, although less pronounced and can be explained by variations in soil moisture and changing photosynthetic activity. However, since at least two independent variables are determining the δ13C value (stomatal conductance and photosynthetic activity which are controlled by the interaction of a variety of environmental parameters) interpretation in terms of environmental change is much more difficult. By using both, hydrogen and carbon isotope ratios, an interpretation in terms of changing δ13C values due to changes in environmental parameters is possible. The response of both isotope ratios on a molecular level to changing environmental conditions can also help to understand the flow of carbon and hydrogen through the different metabolic products of plants. Further studies could include different metabolic products (sugars, acetates, etc.) and follow the flow of carbon and hydrogen to estimate turnover times of these products.

59 Seasonal variations in deciduous tree leaf wax n-alkane δD and δ13C values

5.4.4. Implications for the use of n-alkane δD values as a palaeoclimate proxy For Maple leaves, a change in the relative amounts of the n-alkanes is observed, which make the use of certain n-alkane distributions as biomarkers for specific trees nearly impossible.

However, Beech leaves were found to produce only n-C27. Our results demonstrate that δD values from leaf wax n-alkanes are variable throughout the growing season. Comparison with modelled leaf water δD values suggests that the n-alkane δD values from deciduous tree leaves track the leaf water isotopic composition, integrated over a period of roughly a week by a 2 to 4 week time lag. This suggests, that n-alkanes are not produced by recent assimilates (such as sugars) but are conversion products of other compounds, possibly acetates and alcohols, which are the first compounds appearing in the leaf wax layer. The isotopic signal of the leaf water will therefore be transferred to the n-alkanes after a certain time lag. Since relative humidity is the main factor influencing the leaf water δD value, fossil leaf wax n-alkanes from sediments could be used to reconstruct changes in relative humidity. However, as evident in the lower autumn n-alkane δD values and the low soil moisture content, the soil water availability can significantly alter the modelled leaf water signal. Future models describing leaf water enrichment should therefore include a soil water model. Since most leaves from deciduous trees, incorporated into the sediments, are, at least in the temperate regions of the Earth, autumn leaves, the recorded humidity signal will be the autumn signal. However, when investigating long-term changes of relative humidity over time periods of several thousands of years, this should be of minor importance. In small lakes, where the meteoric water is the source water for the lake and the surrounding vegetation, the difference between aquatic biomarker δD values (for instance the algal derived n-C17 alkane) and leaf wax n-alkanes could serve as a proxy for changes in leaf water enrichment and therefore relative humidity. However, it is not clear yet, if different tree species or even different plants (grasses etc.) show different behaviour. Our data from the Maple leaves show, that significant differences in the different n-alkane δD values exist. Clearly, further studies applying this approach to other tree species, grasses etc. under different climatic conditions are necessary to calibrate this new proxy to relative humidity. However, when investigating one site and by the use of multiple proxies giving additional information on the vegetation (such as pollen), reasonable estimates of changes in relative humidity should be possible.

60 Concluding Remarks

6. Concluding Remarks

In this thesis the major open questions regarding the use of compound-specific hydrogen isotope ratios of sedimentary n-alkanes as a palaeoclimate proxy were resolved. Initial studies, including laboratory experiments (HUANG et al., 2002; SAUER et al., 2001; SESSIONS et al., 1999) have shown, that the δD values of organic compounds, such as sterols and fatty acids seem to reliably record the source water δD value. However, no study before has investigated, if sedimentary n-alkanes, maybe the most abundant biomarker throughout the geological time with no exchangeable hydrogen, record the source water δD signal in a natural system. n- Alkanes also permit a differentiation of aquatic and terrestrial organic matter input into sediments, since different n-alkanes are produced by aquatic and terrestrial organisms. Also, n- alkanes are relatively easy to extract and purify, compared to other biomarker compounds. δD values of these compounds should therefore hold information on the hydrogen source of terrestrial and aquatic organisms. Algae and photosynthetic bacteria living in the water column, which produce the short-chain n-alkanes like n-C17, use the lake water as their hydrogen source, whereas precipitation delivers hydrogen to terrestrial plants, whose leaf waxes contain high concentrations of the long-chained hydrocarbons n-C25 to n-C33. The combined use of aquatic and terrestrial biomarker δD values in closed lake systems should therefore provide additional information on the evapotranspiration of the ecosystem around the lake (Figure 17). In the third chapter of this thesis it was shown that n-alkanes from recent lake sediments along a climatic gradient from Northern Finland to Southern Italy record the source water δD value with high fidelity. The mean hydrogen isotope fractionation during n-alkane biosynthesis for the aquatic n-alkanes in natural ecosystems is constant at -157‰, regardless of lake trophic state or climate, which matches with laboratory experiments (SESSIONS et al., 1999). This result presents a solid base for the application of aquatic n-alkane δD values to reconstruct variations in the hydrogen isotope ratio of the source water, and therefore climate. Furthermore, the n-alkanes of terrestrial origin were enriched in deuterium relative to the aquatic n-alkanes by between 10 and 60‰. This enrichment is not due to a different isotopic fractionation (ε) during n-alkane biosynthesis, but due to evapotranspiration processes in the plants leaves. So, the apparent isotope fractionation between meteoric water and n-alkanes for terrestrial plants is therefore smaller, than compared to aquatic organisms (Figure 17). However, the question remained, if the terrestrial n-alkane δD signal in these lake sediments is a mean annual signal or if n-alkanes in the leaf waxes are synthesised de novo year round. Therefore plant biomass, mostly leaves from the dominating deciduous trees at the lake shore was analysed. The results presented in

61 Concluding Remarks chapter 4 show, that the δD values of the plant biomass n-alkanes are in general more enriched than the sedimentary n-alkanes of terrestrial origin. These data suggest that n-alkanes from the leaf waxes record seasonal variations, since the sampling took place in summer, when evapotranspiration is expected to enrich the leaf water in deuterium, compared to the autumn, when most leaves will be deposited into the lake sediments. The results also reveal that δD values of the leaf wax n-alkanes are similar for different deciduous tree species at the same site, but variable even between the same species at neighbouring sites. This suggests, that changes in terrestrial plant n-alkane δD values record variability in local climate, such as changes in relative humidity. Furthermore, it was observed that with increasing average chain length of the n- alkanes from the leaves of deciduous trees, the hydrogen isotopic fractionation between source water (ε) and n-alkanes increases. This increase of ε is due to decreasing enrichment of leaf water, due to evapotranspiration. Therefore, broadleaf trees with longer chain n-alkanes are better protected from evaporative water loss. n-Alkanes from plants without stomata, such as mosses, are in general more enriched than n-alkanes from deciduous tree leaves at the same site, indicating that stomata play an important role in regulation the water loss of plants.

Figure 17: Isotopic relationships between meteoric water and aquatic (n-C17, n-C19, n-C21) and terrestrial n- alkanes (n-C25, n-C27, n-C29, n-C31) in lake sediments

To assess the magnitude and timing of possible seasonal variations in leaf wax n-alkane δD values, leaf samples from a Maple and Beech stand in Central Germany were sampled and analysed over the 2004 growing season. Both deciduous tree leaves show significant variations

62 Concluding Remarks in n-alkane δD and, to a smaller extend, δ13C values over the growing season. It is well established, that for instance δD values of cellulose from tree rings can record seasonal variations in leaf water (BARBOUR et al., 2002). These findings have resulted in a number of attempts to model the leaf water enrichment as a function of source water and environmental parameters (GAN et al., 2003; RODEN et al., 2000; RODEN and EHLERINGER, 1999). Observations and models have shown, that the source water and water vapour isotope composition and relative humidity are the main drivers for changes in leaf water isotope enrichment. For the Beech stand, the Hainich forest, meteorological data are monitored throughout the year. Therefore, the deuterium enrichment of leaf water was modelled over the 2004 growing season. Comparison of the modelled leaf water δD values with the leaf wax n- alkane δD values revealed a 2-4 week time lag between isotope excursions in modelled leaf water and n-alkane δD values. Studies on the development of the wax layer of leaves have shown, that after an artificial wax removal a complex biosynthesis takes place to re-establish the protective layer. First aldehydes are formed and become dominant in the wax, after 10 days they are converted into alcohols and only after 30 days n-alkanes become the dominant compound in the wax. These results support our isotope data and imply that n-alkanes are not synthesised by recent assimilates, but are conversion products of earlier formed compounds, such as alcohols.

These results highlight the amount of information preserved in n-alkane δD values and open new perspectives in reconstructing changes in the water cycle. In a recent study Radke et al. (2005) could show, that high temperatures during sediment burial alter the n-alkane δD value in Permian and Jurassic sediments. However, since they observed a linear relationship between sediment maturity and n-alkane δD values, a correction of δD values from mature sediments seems possible, if the burial history of the sediments is known. This would extend the use of compound-specific hydrogen isotope ratios to reconstruct palaeoclimate and palaohydrology to periods in Earth history where only few proxy data exist.

It is expected, that δD values of n-alkanes and other biomarker compounds will be increasingly employed to elucidate the influence of past climate change on the water cycle and vice versa. Based on the results from this thesis several perspectives for future applications in this emerging field become possible:

63 Concluding Remarks

(1) The connection of global climate change and the global water cycle in the recent geological past The dominant mode of global climate variability on multi-year timescales, the El Nino Southern Oscillation (ENSO) phenomenon affects the whole climate system, since global precipitation patterns are altered during ENSO events. Worrisome is the prediction of an increase in frequency and/or intensity of ENSO events in a warmer climate (IPCC, 2001). However, there is no agreement among palaeoclimatologists on the magnitude of change in frequency or intensity of ENSO during the last ice age, a time when the climate state was significantly different from today. There also is evidence that the monsoonal cycles in Asia have changed several times abruptly since the last deglaciation, but the timing and intensity of some of these episodes remain elusive (MORRILL et al., 2003). The extact timings, frequencies and intensities and therefore causes and feedbacks of these major climate phenomena, by their nature closely connected to the water cycle, are not well understood. Compound-specific δD values of sedimentary biomarkers from lake sediments in the affected areas can offer new information on the variability of these climatic phenomenons and elucidate the causes of these changes and therefore permit better predictions of the consequences of current human interference with the climate system.

(2) Application of compound specific δD values for plant-water studies The growing interest in understanding plant-water relationships in ecological studies demands for a detailed comprehension of the pathways and storage capacities of water in different plants. By investigating the flow of hydrogen isotopes through the different metabolic products in plants, this new analytical approach can deliver substantial information on the hydrogen and therefore water flow in plants. As the results presented in Chapter 5 show, n-alkanes in leaf waxes are conversion products of earlier formed substances. Future studies could follow the hydrogen flow into different metabolic compartments, to better understand the order in which metabolites are formed and to explain, at which stages in lipid biosynthesis the major isotope fractionation steps take place. Also, these studies could clarify, if hydrogen isotopes in different metabolic compartments record the meteorological conditions over different timeframes. This would provide a tool to reconstruct the prevalent meteorological conditions experienced by the plant.

(3) Determining changes in ecosystem hydrology In a recent study Krull et al. (2005) could show, that the combined use of compound- specific δD and δ13C analysis of n-alkanes from soil organic matter can be used to

64 Concluding Remarks successfully differentiate between water sources of different vegetation types (grasses versus trees) in natural ecosystems. These results show the potential of n-alkane δD values to trace different water use strategies of plants. In the context of the increasing shortage of water in the arid regions of the Earth, understanding of these processes is particularly important in ecosystems altered by humans, such as deforestation or agriculture.

65 Summary

7. Summary

The aim of this thesis was, to explore if the hydrogen isotope ratio of biomarker compounds in lake sediments can be used as a novel tool to reconstruct changes in the water cycle over geological timescales. Therefore surface sediments from small lakes, as well as plant biomass from the lake catchments along a climatic gradient from Northern Finland to Southern Italy were analysed to test, if modern climate variability is recorded in their n-alkane δD values. To clarify the origin of the dominating terrestrial n-alkane input into the lakes and to account for possible seasonal variations in leaf wax n-alkane δD values and their causes, leaves of Beech and Maple trees from two sites in Germany were sampled over one growing season. The results presented here demonstrate various climatic informations are preserved in biomarker δD values. In particular, this new proxy allows the reconstruction of changes in the isotopic composition of lake water, precipitation and relative humidity, all of which are induced by climatic changes. In particular, the following results, related to the questions asked at the beginning (see 1.2.), represent a solid base for the application of n-alkane δD values for palaeoclimate reconstruction:

1) n-alkane δD values from surface sediments of small ground-water fed lakes along a climatic gradient have been shown to reliably record the source water δD values. N-

Alkanes of aquatic origin (n-C17) clearly record the source water δD signal and therefore permit the reconstruction of changes in lake water δD values and hence variations in climate.

2) The isotopic fractionation between source water and n-alkanes in a natural ecosystem of aquatic organisms was found to be constant at -157‰ along the climatic gradient, which

matches results from laboratory studies (SESSIONS et al., 1999). Since the investigated lakes are characterized by different trophic states and cover different climates, we conclude that environmental parameters exert only minor influence on this biochemical fractionation. This is particularly important, when applying n-alkane δD values to palaeoclimate questions. However, the observed constant isotopic fractionation facilitates the use of n-alkane δD values to reconstruct the source water δD value.

3) n-Alkanes of terrestrial origin found in the investigated lake sediments are enriched in D by 10-60‰, due to the evaporative enrichment of D in leaf water. Higher δD values

66 Summary

were observed for n-Alkanes extracted from leaves and plant biomass of the dominating vegetation around these lakes. This suggests that the higher evaporation during the summer (when leaves were sampled) lead to a D enrichment of leaf water and n-alkanes. With increasing average chain length of the n-alkanes from deciduous leaves a larger isotopic fractionation between source water an n-alkanes was observed, suggesting that longer chain n-alkanes in the wax layer of leaves provide a better protection from evaporative water loss and therefore decrease the enrichment of leaf water in plants.

4) If both, terrestrial and aquatic organisms use the same water source, the meteoric water, the observed isotopic difference between both biomarkers in lake sediments can be used to reconstruct leaf water enrichment, which in turn is determined by evapotranspiration. In small, ground water fed lakes, the difference between aquatic and terrestrial n-alkane δD values, can therefore be used as a proxy for the amount of evapotranspiration of the ecosystem at the lake shore.

5) n-alkanes from Maple and Beech leaves show significant seasonal variation in δD as well as δ13C values. Comparison with modelled leaf water δD values from the Hainich site reveals a 2-4 week time lag between isotope excursions in modelled leaf water and leaf wax n-alkane δD values. These observations provide evidence that n-alkanes are not synthesised by recent assimilates, such as sugars, but are conversion products from earlier formed substances such as alcohols. Since air relative humidity is the main driver for changes in leaf water δD values, we conclude that leaf wax n-alkanes reliably track the leaf water δD value and therefore relative humidity. Since most leaves, at least in the temperate climates of the Earth, will be deposited into the lake sedimentary archives in autumn, the preserved δD signal will be the autumn signal.

These results show that compound-specific hydrogen isotope ratios from sedimentary archives record the current climatic conditions of the studied lake ecosystems and hence can be used to reconstruct changes in the water cycle.

67 Zusammenfassung

8. Zusammenfassung

Das Ziel der vorliegenden Dissertation war, zu überprüfen, ob sich substanzspezifische Wasserstoffisotopenverhältnisse (δD) einzelner Biomarker Substanzen aus Seesedimenten als neuer Ansatz zur Rekonstruktion von Änderungen des Wasserkreislaufes in der geologischen Vergangenheit nutzen lassen. Dazu wurden rezente Sedimente aus Seen entlang eines Klimagradienten von Nordfinnland bis Süditalien sowie Pflanzenbiomasse aus deren Einzugsgebieten analysiert, um zu überprüfen, ob die gegenwärtige Klimavariabilität in den δD Werten der n-Alkane aufgezeichnet wird. Um die Einflussfaktoren auf den δD Wert der n- Alkane terrestrischen Ursprungs in den Sedimenten zu bestimmen, wurden zusätzlich Blätter von Buche und Ahorn an zwei Standorten in Deutschland über eine Wachstumsperiode beprobt und analysiert. Die vorliegenden Ergebnisse zeigen die Bandbreite der klimatischen Informationen, die in den δD Werten von n-Alkanen aufgezeichnet werden können und ermöglichen die Anwendung dieses neuen Paläoklimaindikators zur Rekonstruktion der Isotopenzusammensetzung des Seewassers als auch des Niederschlages sowie der relativen Luftfeuchte. Die Rekonstruktion von Änderungen in diesen Parametern erlaubt Rückschlüsse auf klimatische Variationen. Die folgenden Ergebnisse dieser Dissertation bilden eine umfassende Grundlage für die Anwendung substanzspezifischer δD Werte von n-Alkanen als neuer Paläoklimaindikator:

1) δD Werte von n-Alkanen aus rezenten Oberflächensedimenten grundwassergespeister Seen entlang eines klimatischen Gradienten zeichnen die Wasserstoffisotopenzusammensetzung der Wasserstoffquelle von Organismen auf. n-

Alkane die von Algen produziert werden (n-C17) können somit genutzt werden, um die Veränderungen in der Isotopenzusammensetzung des Seewassers, und somit des Klimas, zu rekonstruieren.

2) Es konnte gezeigt werden, dass die Wasserstoffisotopenfraktionierung (ε) zwischen Wasserstoffquelle (meteorisches Wasser) und n-Alkanen in natürlichen Ökosystemen konstant -157‰ über den gesamten Klimagradienten beträgt. Dieses Ergebnis bestätigt Laboruntersuchungen. Da sich die untersuchten Seen in verschiedenen Klimazonen befinden und durch unterschiedliche Trophiestufen gekennzeichnet sind, können Umweltfaktoren, wie z.B. Temperatur als Steuerungsfaktoren für die Isotopenfraktionierung während der n-Alkan Biosynthese ausgeschlossen werden.

68 Zusammenfassung

Diese konstante Fraktionierung ermöglicht die Rekonstruktion der Isotopenzusammensetzung der Wasserstoffquelle anhand der δD Werte von n-Alkanen aus Sedimenten.

3) n-Alkane terrestrischen Ursprungs, die in den Blattwachsen, vor allem von Laubbäumen synthetisiert werden, sind in den Seesedimenten um 10-60‰ angereichert in Deuterium, hervorgerufen durch die Evapotranspiration in Blättern, die das schwere Isotop im Blattwasser anreichert. n-Alkane aus Pflanzenbiomasse vom Ufer der untersuchten Seen zeigten nochmals schwerere δD Werte als die sedimentären n-Alkane. Grund dafür ist eine höhere Evapotranspiration im zur Zeit der Probennahme im Sommer, relativ zum Herbst, wenn die meisten Blätter in die Seen eingetragen werden. Außerdem wurde bei längerer durchschnittlicher Kettenlänge der n-Alkane eine größere Isotopenfraktionierung relativ zum meteorischen Wasser beobachtet, und damit eine geringere Anreicherung von Deuterium im Blattwasser der Pflanzen. Diese Ergebnisse zeigen, dass längerkettige n-Alkane einen effektiveren Schutz vor Evapotranspiration bieten.

4) Wenn terrestrische und aquatische Organismen dieselbe Wasserstoffquelle, das meteorische Wasser, haben, wie beispielsweise an den untersuchten grundwassergespeisten Seen, kann der isotopische Abstand zwischen n-Alkanen terrestrischer und n-Alkanen aquatischer Herkunft zur Rekonstruktion der Deuterium Anreicherung im Blattwasser und somit als Maß für die Evapotranspiration des Ökosystems um Seeufer genutzt werden.

5) Es konnte gezeigt werden, das δD als auch δ13C Werte von n-Alkanen aus Buchen- und Ahornblättern starke Schwankungen über eine Wachstumsperiode aufweisen. Der Vergleich mit modellierten Blattwasserisotopenverhältnissen zeigt, das das δD Signal des Blattwassers erst nach 2-4 Wochen in den δD Werten der n-Alkane erscheint. Diese Ergebnisse zeigen, dass n-Alkane in Blattwachsen nicht direkt aus den Assimilaten der Photosynthese gebildet werden, sondern aus anderen, zuerst gebildeten Metaboliten, wie z.B. Alkohlen. Da die relative Luftfeuchtigkeit der Hauptsteuerungsfaktor für Variationen im δD Wert des Blattwassers ist, kann geschlussfolgert werden, dass n-Alkane der Blattwachse das Isotopensignal des Blattwassers und somit die relative Luftfeuchtigkeit aufzeichnen. Da

69 Zusammenfassung

Blätter zum größten Teil im Herbst in Seen eingetragen werden, ist das im Sediment enthaltene Isotopensignal das der relativen Luftfeuchtigkeit des Herbstes.

Die vorliegenden Ergebnisse bilden die Basis für die Anwendung dieser neuen Methode zur Klimarekonstruktion und zeigen, wie vielfältig δD Werte von n-Alkanen aus Sedimenten zur Rekonstruktion von Veränderungen im Wasserkreislauf genutzt werden können.

70 Acknowledgements

Acknowledgements

Many people have provided help and support during my work for this PhD thesis. Especially I want to acknowledge the help from my advisors at Friedrich-Schiller University Jena and Max- Planck Institute for Biogeochemistry: Dr. Gerd Gleixner (MPI-BGC Jena) for offering me to work in this emerging field of organic biogeochemistry and his strong support during the work for my thesis. I would like to thank my university supervisor Prof. Reinhard Gaupp (Friedrich Schiller University Jena) for his continued support.

I would like to thank my friends and colleagues at MPI-BGC for providing an excellent research atmosphere and support throughout my thesis’ work: Ansgar Kahmen and Volker Hahn for their friendship and fruitful discussions on drafts of the manuscripts. Steffen Rühlow and Jens Radke for the help while sampling and for teaching me how to run a GC-IRMS system. The IsoLab at MPI-BGC, especially Willi Brand, Roland Werner and Heike Geilmann for the measurement of water samples and standards. The workshop at MPI- BGC, especially Bernd Schlöffel and Frank Voigt are acknowledged for construction of our sampling raft.

Without help from the following people the 2002 lake sampling would not have been possible: I am grateful to Gerhard Daut (University of Jena), Jörg Negendank and Cathrin Brüchmann (GFZ Potsdam) for providing sampling equipment. Dan Hammarlund (University of Lund), Tuomas Laurilla (Finnish Meteorological Institute, Sodankylä), Lauri Arvola (Lammi Biological Station), Timo Vesala (Helsinki University), Niels-Otto Jensen (Risø National Laboratory), Günther Schleser and Andreas Lücke (FZ Jülich) and Günther Seufert (JRC, San Rossore) are acknowledged for providing help with selection of the lakes, support on-site and data on the lakes.

I thank Ines Mügler and Andrej Thiele for their help with sample extraction and measurement and Alexander Telz for leaf sampling at the Hainich. I would like to thank the library, the computer department and the administration of MPI-BGC for their support.

71 Acknowledgements

I would like to thank Alex Sessions (CalTech, Pasadena, USA) and Julian Sachs (MIT, Cambridge, USA) who gave very helpful comments on chapter 3; Evelyn Krull (CSIRO, Glen Osmond, Australia), Graham Farquhar (RSBC, Canberra, Australia) and Jim Ehleringer (University of Utah, Salt Lake City, USA) for their comments on Chapter 5.

I also acknowledge the Deutsche Forschungsgemeinschaft for funding my PhD position under project Gl262/8. I would like to thank the Max Planck Society for providing the excellent scientific infrastructure at MPI-BGC.

I am deeply grateful to my parents Sabine and Konrad Sachse for their love, support and encouragement throughout my entire career. I would like to thank my girlfriend, Luzie Polatzek, for her enduring support during work for this thesis.

72 References

9. References

Abelson P. H. and Hoering T. C. (1961) Carbon Isotope Fractionation in Formation of Amino Acids by Photosynthetic Organisms. Proceedings of the National Academy of Sciences of the United States of America 47(5), 623-&. Allen J. R. M., Brandt U., Brauer A., Hubberten H. W., Huntley B., Keller J., Kraml M., Mackensen A., Mingram J., Negendank J. F. W., Nowaczyk N. R., Oberhansli H., Watts W. A., Wulf S., and Zolitschka B. (1999) Rapid environmental changes in southern Europe during the last glacial period. Nature 400(6746), 740-743. Andersen N., Paul H. A., Bernasconi S. M., McKenzie J. A., Behrens A., Schaeffer P., and Albrecht P. (2001) Large and rapid climate variability during the Messinian salinity crisis: Evidence from deuterium concentrations of individual biomarkers. Geology 29(9), 799-802. Baas M., Pancost R., van Geel B., and Damste J. S. S. (2000) A comparative study of lipids in Sphagnum species. Organic Geochemistry 31(6), 535-541. Barbour M. M. and Farquhar G. D. (2000) Relative humidity- and ABA-induced variation in carbon and oxygen isotope ratios of cotton leaves. and Environment 23(5), 473-485. Barbour M. M., Walcroft A. S., and Farquhar G. D. (2002) Seasonal variation in delta C-13 and delta O-18 of cellulose from growth rings of Pinus radiata. Plant Cell and Environment 25(11), 1483-1499. Bowen G. and Revenaugh J. (2003) Interpolating the isotopic composition of modern meteoric precipitation. Water Resources Research 39(10), 9-1 - 9-13. Burgoyne T. W. and Hayes J. M. (1998) Quantitative production of H-2 by pyrolysis of gas chromatographic effluents. Analytical Chemistry 70(24), 5136-5141. Chikaraishi Y. and Naraoka H. (2003) Compound-specific delta D-delta C-13 analyses of n- alkanes extracted from terrestrial and aquatic plants. 63(3), 361-371. Collister J. W., Rieley G., Stern B., Eglinton G., and Fry B. (1994) Compound-Specific Delta- C-13 Analyses of Leaf Lipids from Plants with Differing Carbon-Dioxide Metabolisms. Organic Geochemistry 21(6-7), 619-627. Craig H. (1961) Isotopic Variations in Meteoric Waters. Science 133(346), 1702-1703. Craig H. and Gordon L. (1965) Deuterium and oxygen 18 variations in the ocean and the marine atmosphere. In Proceedings of a Conference on Stable Isotopes in Oceanographic Studies and Paleotemperatures (ed. E. Tongiori), pp. 9-130. CNR- Laboratorio di Geologia Nucleare. Cranwell P. A., Eglinton G., and Robinson N. (1987) Lipids of Aquatic Organisms as Potential Contributors to Lacustrine Sediments.2. Organic Geochemistry 11(6), 513-527. Damesin C. and Lelarge C. (2003) Carbon isotope composition of current-year shoots from Fagus sylvatica in relation to growth, respiration and use of reserves. Plant Cell and Environment 26(2), 207-219. Dawson T. E. and Ehleringer J. R. (1993) Isotopic Enrichment of Water in the Woody Tissues of Plants - Implications for Plant Water Source, Water-Uptake, and Other Studies Which Use the Stable Isotopic Composition of Cellulose. Geochimica Et Cosmochimica Acta 57(14), 3487-3492. Eglinton G. and Hamilton R. J. (1967) Leaf Epicuticular Waxes. Science 156, 1322-1334. Epstein S., Yapp C. J., and Hall J. H. (1976) Determination of D-H Ratio of Non- Exchangeable Hydrogen in Cellulose Extracted from Aquatic and Land Plants. Earth and Planetary Science Letters 30(2), 241-251. Estep M. F. and Hoering T. C. (1980) Biogeochemistry of the Stable Hydrogen Isotopes. Geochimica Et Cosmochimica Acta 44(8), 1197-1206. Farquhar G. D., Hubick K. T., Condon A. G., and Richards R. A. (1989) Carbon isotope discrimination and plant water-use efficiency. In Stable isotopes in Ecological Research, Vol. 68 (ed. P. Rundel, J. R. Ehleringer, and K. Nagy), pp. 21-40. Springer-Verlag. Farquhar G. D., Oleary M. H., and J. a. (1982) On the Relationship between Carbon Isotope Discrimination and the Inter-Cellular Carbon-Dioxide Concentration in Leaves. Australian Journal of 9(2), 121-137.

73 References

Feng X. H. and Epstein S. (1995) Climatic Temperature Records in Delta-D Data from Tree- Rings. Geochimica Et Cosmochimica Acta 59(14), 3029-3037. Ficken K. J., Li B., Swain D. L., and Eglinton G. (2000) An n-alkane proxy for the sedimentary input of submerged/floating freshwater aquatic macrophytes. Organic Geochemistry 31(7-8), 745-749. Flanagan L. B., Comstock J. P., and Ehleringer J. R. (1991) Comparison of Modeled and Observed Environmental-Influences on the Stable Oxygen and Hydrogen Isotope Composition of Leaf Water in Phaseolus-Vulgaris L. Plant Physiology 96(2), 588-596. Flanagan L. B. and Ehleringer J. R. (1991) Stable Isotope Composition of Stem and Leaf Water - Applications to the Study of Plant Water-Use. Functional Ecology 5(2), 270-277. Gan K. S., Wong S. C., Yong J. W. H., and Farquhar G. D. (2003) Evaluation of models of leaf water O-18 enrichment using measurements of spatial patterns of vein xylem water, leaf water and dry matter in maize leaves. Plant Cell and Environment 26(9), 1479- 1495. Gat J. R. (1996) Oxygen and hydrogen isotopes in the hydrologic cycle. Annual Review of Earth and Planetary Sciences 24, 225-262. Gehre M., Geilmann H., Richter J., Werner R. A., and Brand W. A. (2004) Continuous flow H-2/H-1 and and(18)O/O-16 analysis of water samples with dual inlet precision. Rapid Communications in Mass Spectrometry 18(22), 2650-2660. Gleixner G., Kramer C., Hahn V., and Sachse D. (2005) The effect of biodiversity on carbon storage in soils. In Forest Diversity and Function: Temperate and Boreal Systems, Vol. 176 (ed. M. Scherer-Lorenzen, C. Körner, and E.-D. Schulze), pp. 165-183. Springer- Verlag. Gonfiantini R. (1986) Environmental Isotopes in Lake Studies. In Handbook of Environmental Isotope Geochemistry (ed. P. Fritz and J. C. Fontes), pp. 113-168. Elsevier. Grimalt J. O. and Albaiges J. (1987) Sources and occurrence of C12-C22 n-Alkane distributions with even carbon-number preference in sedimentary environments. Geochimica et Cosmochimica Acta 51, 1379-1384. Gulz P. G. (1994) Epicuticular Leaf Waxes in the Evolution of the Plant Kingdom. Journal of Plant Physiology 143(4-5), 453-464. Han J. and Calvin M. (1969) Hydrocarbon Distribution of Algae and Bacteria, and Microbiological Activity in Sediments. Proceedings of the National Academy of Science of the United States of America 64(2), 436-443. Han J., McCarthy E., Van Hoeven W., Calvin M., and Bradley W. (1968) Organic Geochemical Studies, II. A Preliminary Report on the Distribution of Aliphatic Hydrocarbons in Algae, in Bacteria, and In a Recent Lake Sediment. Proceedings of the National Academy of Sciences of the United States of America 59(1), 29-33. Hayes J. M., Freeman K. H., Popp B. N., and Hoham C. H. (1990) Compound-Specific Isotopic Analyses - a Novel Tool For Reconstruction of Ancient Biogeochemical Processes. Organic Geochemistry 16(4-6), 1115-1128. Helle G. and Schleser G. H. (2004) Beyond CO2-fixation by Rubisco - an interpretation of C- 13/C-12 variations in tree rings from novel intra-seasonal studies on broad-leaf trees. Plant Cell and Environment 27(3), 367-380. Herbst M. (1995) Stomatal Behavior in a Beech Canopy - an Analysis of Bowen-Ratio Measurements Compared with Porometer Data. Plant Cell and Environment 18(9), 1010-1018. Hilkert A. W., Douthitt C. B., Schlüter H. J., and Brand W. A. (1999) Isotope Ratio Monitoring Gas Chromatography/Mass Spectrometry of D/H by High Temperature Conversion Isotope Ratio Mass Spectrometry. Rapid Communications in Mass Spectrometry 13, 1226-1230. Huang Y., Shuman B., Wang Y., and Webb T. (2002) Hydrogen isotope ratios of palmitic acid in lacustrine sediments record late Quaternary climate variations. Geology 30(12), 1103- 1106. Huang Y. S., Shuman B., Wang Y., and Webb T. (2004) Hydrogen isotope ratios of individual lipids in lake sediments as novel tracers of climatic and environmental change: a surface sediment test. Journal of Paleolimnology 31(3), 363-375. IAEA. (2001) GNIP Maps and Animations. International Atomic Energy Agency, Vienna. Accessible at http://isohis.iaea.org. IPCC. (2001) Climatic Change 2001: The Scientific Basis. Cambridge University Press.

74 References

Jacob H. and Sonntag C. (1991) An 8-Year Record of the Seasonal-Variation of H-2 and O-18 in Atmospheric Water-Vapor and Precipitation at Heidelberg, Germany. Tellus Series B- Chemical and Physical Meteorology 43(3), 291-300. Jeffrey S. W. and Humphrey G. F. (1975) New Spectrophotometric Equations for Determining a, B, C1 and C2 in Higher-Plants, Algae and Natural Phytoplankton. Biochemie Und Physiologie Der Pflanzen 167(2), 191-194. Jetter R. and Schaffer S. (2001) Chemical composition of the Prunus laurocerasus leaf surface. Dynamic changes of the epicuticular wax film during leaf development. Plant Physiology 126(4), 1725-1737. Knohl A., Schulze E. D., Kolle O., and Buchmann N. (2003) Large carbon uptake by an unmanaged 250-year-old deciduous forest in Central Germany. Agricultural and Forest Meteorology 118(3-4), 151-167. Krull E. S., Gleixner G., and Sachse D. (2005) Compound-specific d13C and dD analyses of plant and soil organic matter: Effects of vegetation change on ecosystem hydrology. Soil Biology and Biochemistry submitted. Leaney F. W., Osmond C. B., Allison G. B., and Ziegler H. (1985) Hydrogen-Isotope Composition of Leaf Water in C-3 and C-4 Plants - Its Relationship to the Hydrogen- Isotope Composition of Dry-Matter. Planta 164(2), 215-220. Li M. W., Huang Y. S., Obermajer M., Jiang C. Q., Snowdon L. R., and Fowler M. G. (2001) Hydrogen isotopic compositions of individual alkanes as a new approach to petroleum correlation: case studies from the Western Canada Sedimentary Basin. Organic Geochemistry 32, 1387-1399. Maffei M. (1996) Chemotaxonomic significance of leaf wax alkanes in the gramineae. Biochemical Systematics and Ecology 24(1), 53-64. Maffei M., Mucciarelli M., and Scannerini S. (1993) Environmental-Factors Affecting the Lipid-Metabolism in Rosmarinus-Officinalis L. Biochemical Systematics and Ecology 21(8), 765-784. Majoube M. (1971) Oxygen-18 and Deuterium Fractionation between Water and Steam. Journal De Chimie Physique Et De Physico-Chimie Biologique 68(10), 1423-&. Mayr C., Frenzel B., Friedrich M., Spurk M., Stichler W., and Trimborn P. (2003) Stable carbon- and hydrogen-isotope ratios of subfossil oaks in southern Germany: methodology and application to a composite record for the Holocene. Holocene 13(3), 393-402. McCarroll D. and Loader N. J. (2004) Stable isotopes in tree rings. Quaternary Science Reviews 23(7-8), 771-801. Merlivat L. (1978) Molecular Diffusivities of (H2O)-O-1 6 Hd16O, and (H2O)-O-18 in Gases. Journal of Chemical Physics 69(6), 2864-2871. Meyers P. A. (2003) Applications of organic geochemistry to paleolimnological reconstructions: a summary of examples from the Laurentian Great Lakes. Organic Geochemistry 34(2), 261-289. Morrill C., Overpeck J. T., and Cole J. E. (2003) A synthesis of abrupt changes in the Asian summer monsoon since the last deglaciation. Holocene 13(4), 465-476. Northfelt D. W., Deniro M. J., and Epstein S. (1981) Hydrogen and Carbon Isotopic-Ratios of the Cellulose Nitrate and Saponifiable Lipid Fractions Prepared from Annual Growth Rings of a California Redwood. Geochimica Et Cosmochimica Acta 45(10), 1895- 1898. Oerlemans J. (2005) Extracting a Climate Signal from 169 Glacier Records. Science 308(5722), 675-677. Pendall E., Betancourt J. L., and Leavitt S. W. (1999) Paleoclimatic significance of delta D and delta C-13 values in pinon pine needles from packrat middens spanning the last 40,000 years. Palaeogeography Palaeoclimatology Palaeoecology 147(1-2), 53-72. Piasentier E., Bovolenta S., and Malossini F. (2000) The n-alkane concentrations in buds and leaves of browsed broadleaf trees. Journal of Agricultural Science 135, 311-320. Radke J., Bechtel A., Gaupp R., Püttmann W., Sachse D., Schwark L., and Gleixner G. (2005) Correlation between hydrogen isotope ratios of lipid biomarkers and sediment maturity. Geochimica et Cosmochimica Acta, rev. version. Ramrath A. (1997) Laminierte Sedimente des Lago di Mezzano (Latium, Italien) - Limnogeologie und Rekonstruktion von Umweltbedingungen der letzten 34.000 Jahre. PhD thesis, Universität Potsdam.

75 References

Roden J., Lin G., and Ehleringer J. (2000) A mechanistic model for interpretation of hydrogen and oxygen isotope ratios in tree-ring cellulose. Geochimica et Cosmochimica Acta 64(1), 21-35. Roden J. S. and Ehleringer J. R. (1999) Observations of hydrogen and oxygen isotopes in leaf water confirm the Craig-Gordon model under wide-ranging environmental conditions. Plant Physiology 120(4), 1165-1173. Sachse D. (2002) Rekonstruktion von paläohydrologischen Bedingungen durch gleichzeitige Nutzung der δD-Werte von n-Alkanen und δ18O- sowie δ13C-Werten von Karbonaten - Am Beispiel des Staßfurtdolomites (Ca2) der Bohrung Jena 106. Diploma Thesis, University of Jena. Sachse D., Radke J., Gaupp R., Schwark L., Lüniger G., and Gleixner G. (2004a) Reconstruction of palaeohydrological conditions in a lagoon during the 2nd Zechstein cycle through simultaneous use of dD values of individual n-alkanes and d18O and d13C values of carbonates. International Journal of Earth Sciences 93, 554-564. Sachse D., Radke J., and Gleixner G. (2004b) Hydrogen isotope ratios of recent lacustrine sedimentary n-alkanes record modern climate variability. Geochimica et Cosmochimica Acta 68(23), 4877-4889. Sauer P. E., Eglington T. I., Hayes J. M., Schimmelman A., and Sessions A. L. (2001) Compound-specific D/H ratios of lipid biomarkers from sediments as a proxy for environmental and climatic conditions. Geochimica et Cosmochimica Acta 65(2), 213- 222. Schefuss E., Ratmeyer V., Stuut J. B. W., Jansen J. H. F., and Damste J. S. S. (2003) Carbon isotope analyses of n-alkanes in dust from the lower atmosphere over the central eastern Atlantic. Geochimica Et Cosmochimica Acta 67(10), 1757-1767. Scheidegger Y., Saurer M., Bahn M., and Siegwolf R. (2000) Linking stable oxygen and carbon isotopes with stomatal conductance and photosynthetic capacity: a conceptual model. Oecologia 125(3), 350-357. Schimmelmann A., Lewan M. D., and Wintsch R. P. (1999) D/H isotope ratios of kerogen, bitumen, oil, and water in hydrous pyrolysis of source rocks containing kerogen types I, II, IIS, and III. Geochimica et Cosmochimica Acta 63, 3751-3766. Sessions A. L., Burgoyne T. W., Schimmelmann A., and Hayes J. M. (1999) Fractionation of hydrogen isotopes in lipid biosynthesis. Organic Geochemistry 30(9), 1193-1200. Sessions A. L., Jahnke L. L., Schimmelmann A., and Hayes J. M. (2002) Hydrogen isotope fractionation in lipids of the methane-oxidizing bacterium Methylococcus capsulatus. Geochimica Et Cosmochimica Acta 66(22), 3955-3969. Sternberg L., Deniro M. J., and Ajie H. (1984) Stable Hydrogen Isotope Ratios of Saponifiable Lipids and Cellulose Nitrate from Cam, C-3 and C-4 Plants. Phytochemistry 23(11), 2475-2477. Sternberg L. D. L. (1988) D/H Ratios of Environmental Water Recorded by D/H Ratios of Plant Lipids. Nature 333(6168), 59-61. Thompson L. G., Mosley-Thompson E., Davis M. E., Lin P. N., Henderson K., and Mashiotta T. A. (2003) Tropical glacier and ice core evidence of climate change on annual to millennial time scales. Climatic Change 59(1-2), 137-155. von Grafenstein U., Erlenkeuser H., Brauer A., Jouzel J., and Johnsen S. J. (1999) A mid- European decadal isotope-climate record from 15,500 to 5000 years BP. Science 284(5420), 1654-1657. Watts W. A., Allen J. R. M., Huntley B., and Fritz S. C. (1996) Vegetation history and climate of the last 15,000 years at Laghi di Monticchio, southern Italy. Quaternary Science Reviews 15(2-3), 113-132. Werner R. A. and Brand W. A. (2001) Referencing strategies and techniques in stable isotope ratio analysis. Rapid Communications in Mass Spectrometry 15(7), 501-519. Xie S., Nott C., Avsejs L., Volders F., Maddy D., Chambers F., Gledhill A., Carter J., and Evershed R. (2000) Palaeoclimate records in compound-specific δD values of a lipid biomarker in ombotrophic peat. Organic Geochemistry 31, 1053-1057. Yakir D., Deniro M. J., and Rundel P. W. (1989) Isotopic Inhomogeneity of Leaf Water - Evidence and Implications for the Use of Isotopic Signals Transduced by Plants. Geochimica Et Cosmochimica Acta 53(10), 2769-2773.

76 References

Yakir D. and Israeli Y. (1995) Reduced Solar Irradiance Effects on Net Primary Productivity (Npp) and the Delta-C-13 and Delta-O-18 Values in Plantations of Musa Sp, Musaceae. Geochimica Et Cosmochimica Acta 59(10), 2149-2151. Yang H. and Huang Y. S. (2003) Preservation of lipid hydrogen isotope ratios in Miocene lacustrine sediments and plant fossils at Clarkia, northern Idaho, USA. Organic Geochemistry 34(3), 413-423. Yapp C. J. and Epstein S. (1982) Climatic Significance of the Hydrogen Isotope Ratios in Tree Cellulose. Nature 297(5868), 636-639. Ziegler H. (1989) Hydrogen Isotope fractionation in plant tissues. In Stable isotopes in Ecological Research, Vol. 68 (ed. P. Rundel, J. R. Ehleringer, and K. Nagy), pp. 105- 123. Springer-Verlag.

77 Curriculum Vitae

Dirk Sachse Address: Max Planck Institute for Biogeochemistry Jena Hans-Knöll-Str. 10 Born on September 24th 1974 in Halle/Saale, Germany 07745 Jena Germany Parents: Sabine Sachse (*1950), High-school teacher Tel.: +49.3641.576134 Dr. Konrad Sachse (*1949), Chemist Fax: +49.3641.577861 Brother: Carsten Sachse (*1978), PhD student, Biochemist email: [email protected] web: http://www.bgc-jena.mpg.de/~dsachse/ EDUCATION: 6/1993 Abitur at Adolf-Reichwein secondary school, Jena, Germany

1995-1997 Studies in Geology at Institute for Earth Sciences, Friedrich-Schiller University Jena, Germany

10/1997-7/1998 participation in ERASMUS/SOKRATES exchange program, studies in Geology and Spanish language at Universidad de Granada, Spain

8/1998-3/2002 Studies in Geology at Institute for Earth Sciences, Friedrich-Schiller University Jena, Germany

5/2001-1/2002 Diploma thesis at Max Planck Institute for Biogeochemistry Jena Title: Reconstruction of palaeohydrological conditions through simultaneous use of δD values from n-alkanes and δ18O and δ13C values of carbonates

3/2002 Diploma in Geology, grade: 1.1

5/2002-5/2005 PhD Thesis at Max Planck Institut for Biogeochemistry Jena Title: Use of compound specific hydrogen isotope ratios as a new palaeoclimatic proxy Tutors: Prof. R. Gaupp (University of Jena) Dr. Gerd Gleixner (Max Planck Institute for Biogeochemistry Jena)

PROFESSIONAL / RESEARCH EXPERIENCE

11/1993-1/1995 civil service at the hospital of the Friedrich Schiller University Jena

8/1999-10/1999 Internship at Max Planck Institute for Biogeochemistry Jena Topic: Small-scale isotopic variations in carbonates

8-9/2000 Geological mapping of the “Subalpine Molasse and the Helvetic Nappes of Central Switzerland” supervised by Prof. F.Schlunegger (University of Bern, Switzerland)

5/2001-1/2002 Diploma thesis at University of Jena and Max Planck Institute for Biogeochemistry Jena

2001 Research assistant to Dr. Volker Hahn at Max Planck Institute for Biogeochemistry: sampling of soil gases at various European forest sites (CARBOEUROPE project)

5/2002-6/2005 PhD Thesis at Max Planck Institute for Biogeochemistry Jena

COMPLEMENTARY EXPERIENCE

1998-2001 elected member of the student council at Institute for Geosciences of the University of Jena

1998-1999 Tutor for Spanish, British and Italian ERASMUS/SOKRATES exchange students at University of Jena

1998-2001 elected student member of the Institute council at Institute for Geosciences of the University of Jena

9/2003-3/2005 elected PhD student representative for the PhD students of Max Planck Institute for Biogeochemistry

2003-2005 student member of the science executive council at Max Planck Institute for Biogeochemistry

2004 co-organiser of a practical one-week class for graduate students of the University of Jena on “Organic Geochemistry”

3/2005 co-organiser of a workshop for PhD students from 6 Max Planck Institutes on “Earth System Science” at MPI-BGC Jena

LANGUAGES English (fluent), Spanish (fluent), Russian (school knowledge), German (native language)

…………………………………………………..

Selbständigkeitserklärung

Ich erkläre, dass ich die vorliegende Arbeit selbständig und unter Verwendung der angegebenen Hilfsmittel, persönlicher Mitteilungen und Quellen angefertigt habe.

...... Ort, Datum Dirk Sachse